pickytri

Roasted Broccoli

2020-04-03T00:00:00-04:00

I’ve always liked broccoli, well compared to most other vegetables. Roasting it with some garlic really makes a great side dish though, with a little effort it comes out miles ahead of steaming/boiling, so that even people who don’t like broccoli will eat it.

Ingredients

2 lbs broccoli florrets
6 cloves garlic, minced
2 tbps olive oil
kosher salt

Nutrition

Total Calories: 616
Weight: 943g

Directions

Pre-heat oven to 450F
Mix broccoli, garlic and salt in large bowl.
Spread broccoli on 2 baking pans and cook for 10mins or until broccoli is just turning brown.

Notes: Top with parmesean cheese if you want to knock it out of the park.

Step by Step

This is a pretty easy recipe but can really make broccoli shine. Roasting any vegetable is great, once it gets some nice little brown on it the flavor can be so much better than boiling or steaming or raw. It takes a little more time than just boiling, but is worth it unless you are really in a time pinch.

Fairly simple ingredients too - just broccoli, garlic, salt and olive oil.

The first step is to pre-heat the oven to 450F. While the oven pre-heats chop the broccoli into smallish pieces. I like to make them all bite size. You can also use the stems if you want, but make sure to skin the rough outer layer off. Next up mince the garlic.

Then combine the broccoli, garlic, salt and olive oil. I like to do this in a large bowl, but if you don’t want an extra dish you can do it in the roasting pan. Make sure to try and get it all evenly coated - use your hands.

After it’s all mixed spread the broccoli out across 2 baking sheets - you can line with tinfoil to make cleanup easier.

Then simple place them in the oven for about 10mins. The ends of some of the broccoli florets should just be turning brown - that’s how I like mine, you can cook for a little more or less time according to your tastes.

Remove from oven and enjoy!

Marathon Training - Part 1: Planning

2020-01-01T00:00:00-05:00

I’ve always wanted to run the Boston Marathon and the NYC marathon, these are some of the biggest marathons in the world and on many runners’ bucket lists. I had a plan in the back of my head for years to start training “sometime in the future” for the races and finally decided that “sometime” could be now.

The Marathon: A Long Race

The modern marathon has its origins in the legend of Phillippides, an ancient Greek messenger who ran from the battlefield of Marathon to Athens to announce that the Persians had been defeated. Legend has it after he proclaimed the victory he immediately collapsed and died. So of course when the modern Olympics began in 1896 the organizers decided it would be a great idea to have a “marathon race” because it might be popular.

And it became popular, for such a grueling event, there are over 800 marathons are organized world wide each year and approximately 500,000 Americans complete a marathon every year. The origin story and distance enshrine the marathon as a bucket list event for many runners, and even those who don’t aspire to run a marathon have a pretty good idea of the training necessary to complete one, or at least know it is hard.

The race distance is 26.2 miles. Running in general is hard on the body, a high impact activity that is injury prone, piling on miles and trying to do them as fast as you can is not a pleasant experience, in fact most marathon plans don’t ever have the runner complete a 26 mile run before the actual event and many elite athletes only race once or twice per year in order to have time to recover.

I’ve only ever run a marathon once, and it was after a 2.4 mile swim and 112 mile bike ride and did not go so well thanks to blisters. Since then I’ve always thought I’d run one by itself and now about 10 years later I am planning to.

The races

The Boston Marathon

Boston is the worlds oldest marathon, having been run since 1897 (inspired by the modern Olympic marathon race) and is also one of the worlds best known road racing events. It draws about 30-40,000 runners and 500,000 spectators every year. It’s hard to find a runner who doesn’t know about Boston or a marathoner who hasn’t dreamed of running it. BQ is a term in more serious running circles for Boston Qualifier, something that will turn a couple heads.

The NYC Marathon

NYC is the largest marathon in the world, with over 50,000 runners competing and about 2million spectators line the course. The race started in 1970 consisted of repeated racing around Central Park. The race organizers decided to change the course in 1976 for the bicentennial to include all five boroughs of NYC, which was a huge success and what was supposed to be a one time course became the standard route. Held in the fall it is a nice complement to Boston in April.

Qualification

Boston and NYC are notorious among runners for their strict time qualification entry requirements. Not only do you have to run a marathon, you have to run a marathon in a fairly quick time to be able to have the chance to run in Boston or NYC. (There are other entry methods like the lottery and charity slots - but part of my goal is to qualify. Don’t get me wrong, I’ll enter the lottery as well, but I am not counting on it). Below is a table with the time qualifications for the 2020 Boston and NYC marathons (even though I’m aiming for the 2021 races):

AGE	GENDER	NYC time (pace)	BOSTON time (pace)
18-34	male	2:53:00 (6:35/mile)	3:00:00 (6:51/mile)
35-39	male	2:55:00 (6:40/mile)	3:05:00 (7:03/mile)
40-44	male	2:58:00 (6:47/mile)	3:10:00 (7:14/mile)
45-49	male	3:05:00 (7:03/mile)	3:20:00 (7:37/mile)
50-54	male	3:14:00 (7:23/mile)	3:25:00 (7:49/mile)
55-59	male	3:23:00 (7:44/mile)	3:35:00 (8:12/mile)
60-64	male	3:34:00 (8:09/mile)	3:50:00 (8:46/mile)
65-69	male	3:45:00 (8:34/mile)	4:05:00 (9:20/mile)
70-74	male	4:10:00 (9:32/mile)	4:20:00 (9:54/mile)
75-79	male	4:30:00 (10:17/mile)	4:35:00 (10:29/mile)
80+	male	4:55:00 (11:55/mile)	4:50:00 (11:03/mile)

18-34	female	3:13:00 (7:21/mile)	3:30:00 (8:00/mile)
35-39	female	3:15:00 (7:26/mile)	3:35:00 (8:12/mile)
40-44	female	3:26:00 (7:51/mile)	3:40:00 (8:23/mile)
45-49	female	3:38:00 (8:18/mile)	3:50:00 (8:46/mile)
50-54	female	3:51:00 (8:48/mile)	3:55:00 (8:57/mile)
55-59	female	4:10:00 9:32/mile)	4:05:00 (9:20/mile)
60-64	female	4:27:00 (10:11/mile)	4:20:00 (9:54/mile)
65-69	female	4:50:00 (11:03/mile)	4:35:00 (10:29/mile)
70-74	female	5:30:00 (12:35/mile)	4:50:00 (11:03/mile)
75-79	female	6:00:00 (13:43/mile)	5:05:00 (11:37/mile)
80+	female	6:35:00 (15:03/mile)	5:20:00 (12:12/mile)

more details on time qualifications:

NYC
Boston

For Boston earning the qualifying time doesn’t mean you are guaranteed entry, just that you can apply to enter and they take the fastest people, so making the time by a small margin won’t get you in. In recent years the actual cut-off time for entry has been 2-5mins below the official cut offs. For NYC the time has to be run at specific NYC Road Runner races (and the marathon I am planning to run is not one of them).

Plan

Now that I know where I need to go to qualify, time to figure out where I am and how to get there from here. There are a million marathon training plans to choose from but I settled on the 18 week 80/20 running plan because it incorporates heart rate training which I am used to and also is based on the plans that elite athletes use. The plan peaks at about 70 miles/week with a long run of 22 miles and is focused on training 80% in an easy zone (for endurance) and 20% in a hard zone (for speed).

I plan to run a marathon in May to hopefully qualify for next years Boston and NYC races - Yes, you have to qualify in the year before you race. I am looking to run a 2:55 or better, and based on how my race in May goes I will see what other races to do the rest of the year.

Where I Am:

The past few months I have been gradually running more in anticipation and increased my mileage to about 30miles/week (along with biking 4 or so hrs). My longest run in this training cycle has been just over 14 miles in just under 2hrs so I am in a good starting place for a marathon training plan mileage wise. In the next 18 weeks I’ll be able to build up more mileage and by dropping bike workouts I will build up even more mileage.

Speed wise, it’s harder to tell. The best predictor of marathon pace is a recent marathon race. After that it’s a recent half marathon race. Neither of which I have done in the past few years. So the last step before getting started was to run a threshold test to re-assess my training zones (which the training plan is based on).

The threshold test I did was a 30min all out run. I dread these workouts because they are physically and mentally tough, and for this one I happened to have a cold which makes things that much worse. But the weather was great - especially for late December, 50F and sunny and I did feel pretty good once I started running.

Here are some stats from that workout:

You can see once my HR gets up it stays there, which is good, you want a nice flat HR for something like this. You can tell it’s right on the line between the yellow and red areas and those are my old threshold zone settings, and my threshold HR is pretty much the same as it was.

Mile	Duration	HR Avg.
1	5:48	180
2	6:17	190
3	6:33	190
4	6:47	190
5	9:34	171

This pace chart shows some areas for improvement. My average pace for the effort was 6:24, which I was happy with, however you can see my first mile was about a minute faster than my 4th mile (I only got about 4.7miles in the 30mins so the 5th mile includes part of my cooldown). I expected a drop in pace by maybe 10-20seconds/mile (which is still pretty significant) but a minute is a huge difference. With a more consistent pace I could probably have averaged 6:15 or faster. This is something I need to work on during my long runs.

Conclusion

There’s a long road ahead between now and running Boston, or NYC, or even the Long Island Marathon in May. Overall I am happy with my position, I am just starting the marathon training but have an achievable goal, a solid plan, and some areas to focus on improving. I’ll be checking in again after a few months for an update on my training and after the race with how it went.

Potatoes Halves with Guacamole

2019-05-04T20:00:00-04:00

I usually consider baked potatoes a winter food, especially when you start adding toppings, but what happens when it’s starting to get warm out and you want a touch of summer on your baked potatoes? Just fill them with some guacamole of course.

Ingredients

2 lb potatoes
2 tbsp canola oil
kosher salt
black pepper
300g (or more as desired) guacamole see my recipe here

Nutrition

Total Calories: hard to tell becuse scooping the insides out of the potatoes.
Weight: also hard to tell

Directions

Preheat oven to 400F and scrub potatoes clean. Slice in half length-wise and place in salted boiling water for roughly 10-15mins until they become just tender.
If you don’t have guacamole now is a good time to make it.
Drain and dry portatoes. Then scoop out the insides leaving about 1/4” potato rim so they still hold their shape. Set aside the insides (I shredded mine and froze it for future use as tater tots).
Brush the inside and outside of the potatoes with canola oil and sprinkle with salt and pepper. Place on baking sheet and cook in the oven for about 10-15mins until they are crispy.
Scoop guacamole into skins and serve.

Notes: Best served on warm potatoes with fresh guacamole that is cold.

Step by Step

This isn’t a recipe I found somewhere but more of a mashup of a few things because I wanted to try them and didn’t find exaclty what I was looking for - so strap in and see how they turned out (spoiler: they were pretty great but probably could be better).

This all started because I wanted to make potato skins with guacamole and use the insides later for tater tots. Most tater tot recipes call for parboiling, and most potato skin recipes call for baking, so I settled on boiling first then scooping out the inside, then baking.

First step was to preheat the oven, then start boiling some salted water and scrub the potatoes.

Then slice the potatoes in half length wise (and ignore the brown parts because we will be scooping them out anyway.

And then they go into the boiling water until tender.

The idea here was to get the insides ready for tater tots, but that didn’t quite work out. The skin was coming off and the outside of the potatoes were tender while the insides were still pretty hard. It made scooping out the insides (the tough part surrounded by tender skin) difficult and at this point I was very skeptical about how these would turn out.

If I come back and do these again (which seems likely) I will scrap the idea of using the insides as tater tots and probably try baking them first. It will probably lead to crispier skins too.

The next step is to place on a baking sheet and coat with canola oil and lightly salt. Coat both sides, and cook with skin side down.

I have to say they were not much to look at when done cooking.

But add some guac and everything is better.

Overall they were very tasty, but I think by trying to boil them and save the insides for tater tots I sacrificed some crispy texture. Next time I will try baking and just use the insides for a hash or mashed potatoes.

Home Lab - Part I: Server

2019-01-05T00:00:00-05:00

In the last few months I expanded my home lab with a new server (and a new router but that’ll be in another post). I purchased a used enterprise server and started migrating my already hosted services over to it (for example this website) and adding services and experimenting with things to learn and grow and have fun.

If you are a little lost, a home lab is the installation and/or configuration of enterprise hardware/software in the home. People do this for fun, to brush up on job related skills, or experimenting with new technologies. It can be as small as a virtual machine on your computer where you mess around with CentOS or as large as multiple full racks that eat up more power than the rest of the neighborhood. This post will serve as an intro to my home lab and what it does.

Physical Server

It’s easiest to start at the tangible things, which in this case is an actual physical server. Technically a server can be anything that serves data, in my case I bought a used Dell R710, which was top of the line enterprise server hardware back in 2012 and fits my needs nicely, especially budget wise.

When I first got the server, things were a bit messy.

That large grey thing is the server:

Plenty of wires running everywhere.

Then I at least put cleaned up the cables a little bit:

That’s the server on the bottom with networking equipment on the top:

And here is a picture with a refurb computer on the desk. You can also see the UPS battery backup in the lower left with the white light on it:

The battery backup unit, which contrary to popular belief isn’t so I can have my server on when the power is out, is responsible for protecting against corruption of data during a power outage. If a computer is in the middle of writing data when the power fails all the data on the disk can become corrupted. On a server this is Very Bad News. To mitigate this risk a battery backup will keep the computer running for 5-10mins after the power goes out to finish writing data and then power down when it is safe. (there is an added benefit during a brown out where power is only out for a minute or power flashes, the server wont shut down - which can take 5mins to boot back up).

The specs:

Component	Spec	Notes
CPU	2x Intel Xeon X5660 Hex Core 2.8GHz	24 Cores
RAM	128GB DDR3 RAM
HDD	6x3TB Drives	RAID5 (13TB useable)
HDD	1x8TB USB Drive	For Backups
Network	4x1000/100/10 Ports
Network	1x100/10 iDrac	Remote Management

This machine is a beast. It would be wildly overspecced for a typical home user and even a typical small/medium business would not need this much computing power so why do I have it? Virtual Machines.

Virtual Machines

Now you might be asking what the heck is a virtual machine? And why would that require a lot of computing power?

Well, the simple answer is a virtual machine is an emulation of a computer system. It provides the full functionality of a real computer system inside of another one. You assign your virtual machine resources (say 2xCPU cores, 8GB RAM and 50GB Hard drive space) and now you can use that virtual machine as a computer. If you have a particualary beefy physical host machine you can deploy many virtual machines. Inside these virtual machines you can install operating systems, different programs and they will act like a physical computers with the allocated resources (and maybe a slight 5-10% performance hit from being a virtual machine).

So now we have a bunch of virtual machines, but why? I basically described setting up a slower computer inside your computer. Which doesn’t make a whole lot of sense at first, but think about the virtual machine more like a file and you can start to see the possibilities.

I can save this file and load it up on another host machine - making backups or dynamic scaling very easy. Imagine a website that gets very popular at certain times of the year, for example a website that has information on daylight savings time. For most of the year I only need one server, but for the week before and after the clocks change my one server becomes overloaded with everyone looking up information on daylight savings time. With virtual machines I can copy my virtual machine that serves the information onto other machines as needed. I don’t need to buy another physical server that will sit idle most of the year, I can rent a server in the cloud and use my virtual machine on it when needed.

There are many other advantages of virtual machines including security, hardware utilization, image based snapshots/backups, and more. Breifly, security wise, rather than installing multiple services on a single machine (potentially allowing security holes in one service to compromise the other services) you can install services on multiple virtual machines and they are now separate and if one service is compromised the others are unaffected. Hardware utilization also benefits from virtual machines, allocating virtual resrouces as needed is much easier than installing or removing physical hardware - and often times can be done without even turning off the virtual machine. A good hypervisor (virtual machine host) will also incorporate snapshots, which are images of the state of a virtual machine - before upgrading a major service on a virtual machine you can make a snapshot - if the upgrade breaks something or doesn’t work correctly you can roll back to the snapshot.

For me the main advantage is flexibility and cost/space/power savings. I can have 20 virtual servers hosted on one physical server, rather than having 20 physical machines which would require more space, cost more and consume more power. The next biggest advantage for me is the ability to run different operating systems, my linux based host machine can have windows guest virtual machines.

If I read about a new technology I want to test out, for example an image hosting service, or a caching server, or a new web server technology, I can spin up a new virtual machine to test it out. This new virtual machine won’t change or affect my current virtual machines, so it won’t interfere with my webserver for this blog. And if I don’t like it, I can just delete it.

Example Virtual Machines

Currently I have about 30 virtual machines is various states of testing and production. I even have one for monitoring them all.

Below is a screen shot of proxmox (a hypervisor, basically a manager for the virtual machines). On the left is a list of some of my virtual machines and on the right is an overview of the load on the host machine (my server):

To keep a better eye on all these virtual machines and their resource usage I have set up a monitor server (also a virtual machine) to collect and format the data, here is a monitor screen from my home lab showing CPU usage and RAM usage:

Everything starting with VM or CT is a virtual machine (or container, which is like a virtual machine but shares more resources with the host). At a glance I can quickly see average usage on each machine. Something like this would not be out of place in a production Network Operations Center or Data Center to monitor health and status of servers relied upon by people for their online services.

Here is a table of some services I am currently running in virtual machines and an explaination of what they do:

Main Service	Virtual Machine Name	Notes
DNS forwarding	CT-DNS01	Ad blocking on DNS level.
Apt Cache	CT-APTCACHE	Apt is an update package system in linux, this server caches updates locally and serves them to other virtual machine, saving internet bandwidth - in the last 2 months alone it has saved close to 3GB in updates downloads.
Reverse Proxy	CT-RPROXY	A reverse proxy is a target for external requests to be forwarded to internal locations, rather than have each external service have ports open to the internet everything is pointed to the reverse proxy first. Advantageous for security and caching.
Local Web	CT-LOCALWEB	Local web services including network documentation, link page for internal network services.
Netbox	CT-NETBOX	Inventory/IP tracking and management.
External Public Web Server	CT-SERVLIVE-PUB	web server for public services like this website.
External Private Web Server	CT-SERVLIVE-PRIV	web server for “private” services I still want to access on the internet, for example password manager, inventory manager, file sync, kanboard project tracking.
Network Boot Image Server	VM-FOG	network booting and OS image server for network boot and installing OS for refurb computers.
Steam Cache	VM-STEAMCACHE	Caching for steam games.
Media Server	VM-Media	Media server for serving media.
Ansible Test Server	VM-ANSIBLE	Ansible is a configuration management tool, I have created a server to test it.
VPN Server	VM-VPN	VPN for accepting remote connections to access local network resources.
Monitor	VM-MONITOR	monitoring with zabbix and grafana (the dashboard above showing CPU and RAM usage).
IPA	VM-IPA	Currently testing IPA as an authentication server.
GNS3	VM-GNS3	Network emulation tool, also uses virtual machines to emulate network devices like routers and switches for learning network configuration.
Domain Controller	VM-WIN2016-DC01	Windows Domain controller.
Windows Refurb Image	WIN10REFURB	Virtual machine of refurb image install.
Windows Test machines	various	about 5 windows test machines for various windows testing scenarios.

Below are some examples of the services:

This is part of the link page to internal services, generally each button is hosted on a different virtual machine and provides a different service:

An example of the Pi-Hole DNS ad blocking service:

A screen shot of downloading a game from the local steam cache server I set up, note the download speed of 113MB/s. This is about Gigabit speed, if I was downloading from the internet I would be seeing only about 7MB/s:

And below is a screen shot of my desktop with a number of virtual machines open, I outlined the important things to note in green:

The top left window is from a virtual machine running Windows Server 2016 with 8GB of RAM, below that is a Windows 10 virtual machine with 4GB of RAM, to the left of that is a Linux Distribution I was testing out which I allocated 16GB of RAM. Above that are two terminal windows showing the usage of the server (left) and the desktop I am using (right).

Conclusion

As you can see there are lots of things to do with a home server as part of a home lab. On the other side of the home lab is the networking equipment, and virtual LANs which I will save for a later post.

I am very happy with my server, although I would like to get a dedicated NAS (network attached storage) to experiment with more robust backup solutions and faster data storage speeds. And I am already getting close to maxing out my RAM usage… I also want to experiment more with data center level virtualization with multiple hypervisors and remote data centers and high availability containers…

All Day Tomato Sauce

2018-10-13T00:00:00-04:00

If you want to make amazing tomato sauce with deep complex flavors and will be home all day and can stir it every couple hours, this is the recipe for you. It’s another recipe from Kenji and seriouseats and it really out shines most other tomato sauces I’ve had, and really blows the jarred sauces out of the water.

Ingredients

4 28-ounce cans of whole peeled tomatoes
1/4 cup extra virgin olive oil
4 tbps butter
1 head garlic
1 tsp red pepper flakes
1 tbps dried oregano
1 medium carrot cut into large chunks
1 medium onion cut in half
1 large stem fresh basil
Salt and Pepper
1/2 cup fresh basil and/or parsley minced

Nutrition

Total Calories: 1392
Weight: 3285g

Directions

Adjust oven rack to lower position and preheat oven to 300°F. Place tomatoes in a large bow and crush or blend to desired consistency. Reserve 3 cups in sealed container in refrigerator for later.
Heat olive oil and butter over medium heat in a large Dutch oven until butter is melted. Add garlic and cook, stirring, until softened and fragrant but not browned, about 2 minutes. Add pepper flakes and oregano and cook, stirring, until fragrant, about 1 minute. Add tomatoes, carrot, onion, and basil, and stir to combine. Season lightly with salt and pepper. Bring to a simmer over high heat.
Cover Dutch oven with lid slightly ajar and transfer to oven. Cook, stirring once every 1 to 2 hours, until reduced by about half and darkened to a deep red, 5 to 6 hours (reduce oven temperature if the sauce is bubbling too rapidly or the browned bits begin to turn too dark.
Remove from oven. Using tongs, discard onion halves, carrots, and basil stems. Add reserved tomatoes to sauce and stir to combine. Season generously with salt and pepper and stir in minced herbs along with additional olive oil as desired. Serve immediately, or allow to cool at room temperature, transfer to airtight containers, and refrigerate for up to 1 week. Sauce can also be frozen in sealed containers for up to 6 months.

Notes: Serve over pasta, or on pizza.

Step by Step

The only special equipment you’ll need for this recipe is a dutch oven, and dutch ovens are not only for making sauce, I like to make bread in them and they are also great for deep frying. The ingredients are also pretty simple:

The first step is to move the oven racks so the dutch oven can fit in there on the bottom rack and preheat the oven to 300F, and don’t worry if you forget about it, 300F isn’t very hot (for ovens) and only takes a couple minutes. Next you want to get your tomatoes to their desired consistency for the sauce. Kenji suggests using your hands and crushing them, but I like smoother sauce so I put them through the immersion blender. This probably messes with the flavor somehow, but so far I haven’t gotten any complaints.

Next up, reserve 3 cups for later. The first time I made this sauce I forgot this step, and the sauce still came out really good. It wasn’t as fresh tasting, but that might have been in my mind, because I’m not sure I could tell the difference between them on a pizza.

Now you can chop up the garlic (probably the longest part of hands on time for this recipe), slive the onion, carrot and get your dried spices ready.

Heat the olive oil and butter in the dutch oven until the butter is melted and throw in the garlic and saute for a few minutes.

Then add the spices and saute for another minute.

Next you can add the tomatoes, onion, carrot and basil stem. Give it a good mix and bring it up to simmering.

Now place it in the oven with the lid slightly ajar and come back and stir once every hour or two. It will get darker and reduce.

Try to stir the brown bits back into the sauce.

Put the lid back on and place back in the oven for another hour or two. Repeat this process for about 6 hours.

Before you take it out for the last time get your fresh herbs ready

And here it is after about 6 hours in the oven, reduced by about half.

Remove the onion halves, carrot chunks and basil stem (if you can find it). Now you can add the reserved portion of tomatoes from earlier and the minced herbs. Don’t forget a generous helping of salt and pepper. And a good stir and all done.

Stir some more to make sure it is smooth (and you got all the carrot chunks out) and taste to make sure it is the amazing tomato sauce promised.

Enjoy on any foods where tomato sauce is called for.

Solid State Drive

2018-08-08T00:00:00-04:00

One of the best bang for your buck upgrades you can make for your computer is a solid state drive (SSD). Of course it depends on what your computer already has, and how you use it, but for me - when I often restart because I dual boot windows and linux and I was still working off an old platter disk, it made absolute sense. So this post is about my new SSD and how it compares to my old drives.

Traditional Hard Disk Drive (HDD) Vs. Solid State Drive (SSD)

Your typical hard drive used to be a couple of actual platters that stored data. It used magnets and had moving parts (the platters spun around), and was pretty good. For storage you generally want it to be big, fast and cheap, the trade off is you can only pick two of those. Platter hard drives are big and cheap, if you need a bunch of storage and don’t want to break the bank a hard drive is for you.

A traditional hard drive (image from wikipedia), you can see the arm and the platter - each time you access data the platter has to spin around to the correct position.

For a few years now solid state drives have been gaining ground as the standard storage device, especially in smaller devices like phones and laptops. They are smaller, don’t have moving parts, use less power and are much faster than traditional hard disk drives. They are also more expensive. A 4TB hard disk drive is generally a couple dollars cheaper than a 512GB SSD (that’s 0.5TB). So if you want the best of both worlds a mixed setup is the way to go, a SSD for important data that is accessed often and a HDD that stores large amounts of data. That’s my new setup, a SSD to store my operating system so it can boot up fast, and a couple HDDs to store large data files.

So just how much faster are SSDs?

Right after installing my new SSD it was clearly faster. The first thing I measured was boot times, I like to play computer games and I also like to use a linux OS. Most of my games are on windows so I have to restart my computer to play them, and when it takes a few minutes to restart it can get annoying - especially if I feel like playing a game then change my mind after 5mins.

Anyway, here are the measurements, the Boot Menu is a screen where I choose to boot to linux or windows:

Storage Drive	Boot Menu (s)	Login Screen (mm:ss)	Useable Desktop (mm:ss)	Total Boot Time (mm:ss)	Shut Down (s)
Traditional HDD	19.48	02:09.41	01:29.51	03:58.4	24.37
New SSD	19.44	00:09.88	00:12.41	00:41.73	14.70

As you can see, the SSD trounces my old hard drive, it’s even faster at shutting down. And while the SSD is a fresh OS install, I don’t see any program I install taking up another 3 minutes during boot.

And if that doesn’t convince you, I even ran some benchmarks, here is the HDD:

and the SSD:

The pertinent numbers here are the average read speed, average write speed and access time and the SSD is the clear winner by far in all categories. I’ve put these in a table as well:

Storage Drive	Avg. Read (MB/s)	Avg. Write (MB/s)	Avg. Access Time (ms)
Traditional HDD	158.8	106.1	18.39
New SSD	560.7	505.9	.03

The read and write speed of the traditional hard drive is pretty close to 125MB/s, and that is an important number because it is how fast a 1Gbps network can transfer data. Most home networks are 1Gbps (not the connections to the outside world, but the connections between your computers in your home), so when transferring data between my laptop and desktop computer my network will support about 125MB/s. My old hard drive, on average will only be able to read 158MB/s (so if I am doing something that is using my hard drive that number will start to dip and my hard drive won’t be able to keep up with my network) and the average write is only 106MB/s, already below my network speed. (this isn’t the whole story - there’s RAM and buffers and overhead for network protocols, but it is a good general illustration of why traditional hard drives are becoming the bottleneck in a home network transfer). The SSD is well above the network speed and will have no problem keeping up with large home network data transfers.

Another thing I’d like to point out is the access times, you can really see that the old HDD with it’s moving parts is simply outclassed by the SSD.

And some more notes:

Modern SSDs are becoming so fast that the traditional interface can’t handle the speed. My SSD is not the fastest on the market, it still uses the same old interface that platter disks use. I’ve read that the upgrade from a SSD with the older interface to the newer one is a similar jump from an old HDD to a SSD.
RAID can mitigate the slow read and write times for traditional hard disks (RAID is combining multiple hard drives so they look like one to the computer). I am putting together a server with 6x3TB hard drives, I put them in RAID 0 (the fastest and most unsafe RAID level, one disk dies and all your data is gone) and I saw a read speed of 814MB/s. In RAID 5 (which is a little slower and a little safer) I saw read speeds of 450MB/s.

2010 - SavageMan

2018-07-01T00:00:00-04:00

2010 was a big year for triathlons for me, I did Ironman France in the early Summer, qualified for and competed at duathlon worlds in Edinburgh and did SavageMan just 2 weeks after that (along with a few shorter races here and there).

SavageMan is a half iron distance race (1.2 mile swim, 56 mile bike ride, 13.1 mile run) in Deep Creek Maryland and it’s special because it is hilly, really hilly, including The Wall - a hill at the start of a 7 mile climb with over a 31% grade. That’s steep, so steep that cars aren’t allowed on it, so steep that about half the people that attempt the climb on a bike fall over, so steep that if you don’t fall off your bike and finish the race you get a brick with your name in it paved into the road. After I heard about the brick of course I had to do this race, I mean you can get a brick! What follows is my account of the 2010 SavageMan half Iron distance race along with some [present day comments in brackets].

Race Report

Pre Race

For savageman I was worried about the bike (even though I did IM france this year and du worlds which was a little hilly as well). Mostly because Debi [my coach] suggested getting a different cog in the back to make it easier to climb, which I didn’t do for France so thought I would be alright, until prerace talking to the other racers who pre rode the course (something else I should have done) and they all had switched out their cogs. A few days before the race I also decided not to go for time, but just try and get a brick - which was after my unsuccessful attempt at running without socks which resulted in large blisters.

Swim

I was comfortable going into the swim and to save my legs (and because I never do) I didn’t kick hard. The water was a good temperature for wetsuits - I wasn’t cold and didn’t get hot either. Before the race I put on 2nd skin to try and cover my blisters, the guy at the running store said it would stay on in the water - it did not. The run up to transition was pretty quick and when I got to my bike I debated taking the time to put on more 2nd skin, but then the guy next to me showed up and I decided to just go without.

Bike

[it was beautiful out in Deep Creek]

The bike starts with a small climb out of the state park and then about 18miles of downhills and small rollers. It got pretty cold because I was barely pedalling and flying downhill at 30mph and still all wet from the swim. Then we get to Westernport where there is a short gradual uphill, a lefthand turn and then you see The Wall. At each cross street it went flat and then seemed to get even steeper after that, and from the bottom it is hard to believe a bike wouldn’t just flip over backwards when trying to go up. I went up the first section in my 2nd to lowest gear and switched to the lowest when I hit the first cross street (could have used that easier cog here), but I stood up and just cranked it out.

[it’s hard to show slope in a picture, but trust me it was steep]

By the second to last section I was vaguely aware of my surroundings, before the race I made a mental note to look at the street for bricks from previous years and I did not find them. The last section has broken pavement which did not help, but also crowds and crowds of people, and somehow through all their cheering and shouting I made it all the way up. Of course then there was another 6 miles of climbing! The rest of the course I took it easy on the downhills/flats and then shot into z4-5 on the named climbs while not trying to go fast, just trying to get over them. Blisters were doing ok since the heel of my cycling shoe did not reach them.

[I’m the second rider in the video, the race organizers were supposed to have a great camera setup and get everyone going up The Wall, but it malfunctioned race day. Luckily I was able to find this clip of me from some random person on youtube who was watching the race]

Comparing this bike course to France, France had longer climbs but nothing as steep and also longer descents which were really nice. I averaged 16.9 at savageman and 17.7 at France so it is a slower course even though its half as long. The scenery in France is not outdone by Deep Creek, it was very nice in Maryland but I didn’t want to get off my bike and just take it all in (although I did want to get off my bike and give my legs a rest). I’m not sure if its because I can still feel the soreness of savageman (and by this point France is a happy memory) but I think it was a tougher course than France. If I had to choose one bike course to do again it would be France.

Run

The run is a 2 loop course with rolling hills and about 3/4 into each lap a gravel trail climb for about 1/4 of a mile. First loop was quicker than the second, and I could feel myself slowing down even when I hit the 10 mile mark. I made it all the way up the trail climb both times and took my first walking stint at the top on the second lap. Told myself it was all downhill from here and finished strong. Blisters hurt a bit during this run (and I got a few new ones) and my racing flats now have some nice dried blood on the inside - but the shoes did pretty good considering.

Post Race

Overall I was very pleased just to finish, and to not fall off my bike. After showering and getting some food my mom came over and told me I got third in my age group, I didn’t believe her. I went and checked the results myself and still didn’t really believe it. But now I have a new decoration for my office!

Results

Overall	5:45:54
Swim	29:32 (1:31/100m)
T1	3:33
Bike	3:18:15 (16.9mph)
T2	2:13
Run	1:52:23 (8:34/mile)

FOG Project

2018-05-27T00:00:00-04:00

The Fog project is an image server I setup to help out when deploying new (well in my case old computers). I started volunteering to refurbish donated old computers for people who otherwise can’t afford one. The steps of refurbing each computer include installing windows 10 and a few programs, which while not all that difficult can be time consuming. An image server can really cut down on install time.

Hard Drive Images and Image Servers

To understand what exactly I set up (and why) I first have to explain a bit about hard drive images, image servers and network booting. Don’t worry if you don’t understand something, the main takeaway should be it is a massive time saver for a little up front effort.

Hard Drive Images

Hard drives are what store all the programs and files on your computer, an image of the hard drive is a snapshot of the current state of the drive (almost like a list of all the 1’s and 0’s on the drive). You can clone images to create exact copies of the hard drive. The ability to copy the state of the drive is very useful in a number of situations, for example backups, if I take an image of a server every day at midnight and something happens to the server - as long as I have that image, I can recreate the state the computer was in when I took the image. Even if the server was destroyed because the building was flooded and covered in water. If the image was safely stored somewhere offsite I can deploy it to another hard drive in another computer and my server will be back online (*some conditions apply - but for the most part it can be that easy).

Similarly if I am a company with 5000 employees and want to get them all working off the same computer programs and settings, I can create an image and deploy it to all the computers, rather than manually install windows, change the settings, and install the required programs on each computer. Or if I was planning to refurbish a number of computers and they all required the same setup, I could create an image and deploy that image to each computer I was refurbishing.

Image Servers

Image servers are servers that are responsible for storing, capturing, deploying and basically managing the hard drive images. The high level work flow is that you first do a manual install of the operating system and programs and get everything just right and then have the image server capture that image, which can then be deployed to different groups of computers. If you were a large company and had different departments like accounting, design and engineering it wouldn’t make sense to deploy the same image to everyone - there are programs that each department needs that will never be used by other departments, so an image server can store multiple images and deploy them to specific groups. Another great thing about images is that it doesn’t matter what OS you are using. Images work on the whole hard drive, so Windows, MAC or Linux doesn’t matter.

So these images sound great, but how are they transferred from the server to the clients? That’s where network booting comes in. When you turn on your computer it looks in certain locations for a program to start running, the most common locations are the hard drive, a USB drive, a CD/DVD-Drive or the network, when it finds a program it will start to run it. You can tell the computer where to look first, so after the network image server is up and running you set the network boot priority to the highest level on the client and then (if things are set up correctly) the client will load the network boot software from the image server and you can capture or deploy an image.

Another cool thing about image servers, you can deploy an image to a more than one client at the same time - if you have 20 computers to set up, manually going through each one and installing windows, getting the settings right and then installing all the programs can take 2-3hrs for each computer, you are looking at 40+ hours of work. Using an image server, you first have to spend 2-3hrs getting one computer setup, take 20mins to capture it and then the rest will take about 20mins to deploy, so around 8-9hours. Using an image server with multicast, all you have to do is set up the image once (2-3 hrs), capture (about 20mins), and deploy simultaneously to all the other machines in about 20mins and you are all done in less than half a day (I haven’t played around with multicast image deployment yet, but think I might for the next batch of refurb computers I pick up). And if suddenly there are 40 computers, as long as you have the image all it takes is another round of deployments (~20mins).

Deploying an Image

Here is an example where I actually deploy an image from the image server to a client. First I head over to the web interface of the image server to make sure it is up and running:

From the web interface you can view images, schedule a capture/deployment, and do a whole lot more.

Next, the first step on the client is set to network boot, and after you turn it on, it looks for the network boot server:

Some details you might see here are the client gets an IP address of 10.0.0.182, the FOG server is 10.0.0.5. Next up the client boots to the FOG project menu:

You can see I already registered this client (host) as L0001, and there are a number of options including just booting off the hard disk to deploying an image. After selecting deploy image you can choose which image to deploy:

There’s only one choice right now because I deleted the other image captures I made. After selecting an image there will be some checks:

And then it will start copying the image:

The image here is of an 80GB disk, and only using 20GB of space. On the image server it is compressed and takes up about 6-7GB. I’ve deployed a few images and found the biggest bottleneck in the process to be the network, if I am deploying to a virtual machine on my laptop SSD over wifi it will take about 8-10mins and when deploying to a 10 year old computer over an ethernet wire it takes about 2-3mins. You can also see the progress from the web interface:

And it’s pretty real time:

On the left is the virtual machine with the image deployment and on the right is the web interface. After the image is copied the client will restart and if you boot from the hard drive you just copied the image to it will start windows:

And will configure based on your devices:

Windows 10 is a massive improvement over earlier versions in image deployment. In previous versions sometimes deploying an image to a computer with a different motherboard or network adapter would cause failures, microsoft even has a way to generalize an image so you can create a non specific version, however I found using it more trouble than setting it up with Windows 10. So far an image taken from an old HP works on an old Dell and a virtual machine with no changes needed. Here is one of the things we need to install for refurbished computers, a EULA that displays when you first boot the computer:

After the user accepts it won’t show up anymore, but we also install a few programs like Firefox, Chrome, CCleaner and others to make it easier for the user to start using their computer. Oh and we do all the windows updates - which can take quite a while.

This whole deployment took roughly 15mins - which is pretty good comparing it to a manual install.

Real World Time Savings

So hopefully you can now see that even in my volunteer work, even if I only refurbish two or three computers a month saving about 2 hours each computer can really add up. Of course it took a few hours to set up the server, and it will take more time as I test things like multicast deployment, but it should save time in the long run. Unless it becomes something I continue to tweak and mess with.

As there is an XKCD for everything…

Scalloped Potatoes

2018-05-17T00:00:00-04:00

Seriouseats calls these Hasselback Potato Gratin and they are a great potato dish, although they do take some time to make - not ideal for a weeknight meal.

Ingredients

85g (3 ounces) finely grated Gruyère or Comté cheese (I used cheddar)
60g (2 ounces) finely grated Parmigiano-Reggiano cheese
2 cups (480ml) heavy cream
2 medium cloves garlic, minced
1 tablespoon fresh thyme or rosemary or other herb leaves, chopped
Salt and pepper
3 to 3.5 lbs russet potatoes, peeled and sliced to 1/8”

Nutrition

Total Calories: 4237
Weight: 2673g

Directions

Preheat oven to 400°F. Combine cheeses in a large bowl. Transfer 1/3 of cheese mixture to a separate bowl and set aside. Add cream, garlic, and fresh herbs to cheese mixture. Season generously with salt and pepper.
Add potato slices and toss with hands until every slice is coated with cream mixture, making sure to separate any slices that are sticking together to get the cream mixture in between them.
Grease a 2-quart casserole dish and add potatoes in neat stacks with their edges aligned vertically. Continue placing potatoes in casserole until all potatoes have been added. Potatoes will be very tightly packed. Pour excess cream/cheese mixture evenly over potatoes.
Cover tightly with lid/foil and transfer to oven. Bake for 30 minutes. Remove lid/foil and continue baking until top is pale golden brown, about 30 minutes longer. Carefully remove from oven, sprinkle with remaining cheese, and return to oven. Bake until deep golden brown and crisp on top, about 30 minutes longer. Remove from oven, let rest for a few minutes, and serve.

Notes

Takes 1.5hrs to cook, total time about 2-2.5hrs so plan accordingly.
Cream mixture will likely spill over sides, place pan below to catch drippings.

Step by Step

I sometimes have trouble getting recipe timing correct, and especially with this dish. The first time I made it I only skimmed the recipe and saw the first 30min cooking time, not the additional 60mins cooking time. Another time I didn’t peel and slice the potatoes early enough for them to be done cooking when everything else was ready. Waiting on one food to finish while everything else is ready can be a real bummer, so there’s a couple things I do to mitigate timing issues. First is to read the whole recipe, really read it. Make sure you know the total cooking time, add prep time to your plan, and make sure you have all your ingredients on hand.

Next, I like to get all my ingredients out and ready. That way when you are done with one recipe step and ready to move on to the next everything is all set. For example if the recipe calls for frying something for 3 mins then adding a different chopped ingredient, having that chopped ingredient ready to go makes cooking so much easier than quickly trying to find it in the fridge and chop it while the ingredients frying are getting overcooked.

And make sure to chop/measure the ingredients. It’s one thing to have a few garlic cloves or a block of cheese on the cutting board, and another to have minced garlic or grated cheese ready to throw in. This is the reason why some recipes that say “total time 30mins” can end up taking an hour to cook, if all the ingredients are ready it will only take 30mins.

And immediately following my explanation on why it’s a good idea to prep everything before making the food I go against my own advice in this recipe and add two thirds of the cheese to the cream and season with salt, pepper and your fresh herbs. But there is a reason, the sliced potatoes are going straight into this mixture, and having it ready before you slice can prevent the potatoes from oxidizing (getting a reddish tint from being exposed to air for too long).

So put the cream mixture aside and get your mandoline ready (not to be confused with the musical instrument), mandolines are great for quickly making uniform slices of anything you slide down the blade (including fingers so be careful). I’ve tried slicing potatoes by hand before, and even posted the results on this blog, it was not pretty. No worries with a mandoline though, potatoes come out great in no time at all.

After slicing, place potatoes directly into cream mixture.

Mix with your hands, coating each and every slice as best you can. Your hands will get messy.

It’s best to have the casserole dish ready to go before mixing the potatoes, greased and everything, that way you can immediately start stacking the slices after they are coated.

After squeezing them all in the dish (it will be tough) pour the cream mixture over the top.

Cover with a lid or tinfoil and get it ready for the oven. The cream mixture has a tendency to spill over the sides so having a drip pan ready to catch those can save your oven some cleaning. After 30mins remove the lid and put back into the oven.

After another 30mins top with the cheese you set aside earlier and put it back in the oven.

And finally after another 30mins it will be just about ready. The top should be dark golden brown.

I probably would have cooked these another 10mins or so. They came out really good, but another few minutes and all the potatoes would have had a nicer crunchy top rather than just most of the potatoes.

Artisan Bread

2018-05-03T00:00:00-04:00

This is a quick (relatively because it still takes about 5-6 hours) white bread recipe from seriouseats. I only do half the recipe and ignore some rising/folding steps, but it comes out really good anyway.

Ingredients

500g all purpose flour
350g room temp. water
11g salt
2 grams instant yeast
vegetable, canola oil for greasing

Nutrition

Total Calories: 1820
Weight: 850g

Directions

Combine flour and water in bowl and let rest for 15-60mins.
Add salt/yeast and mix until incorporated and dough is smooth, move to lightly oiled bowl, cover and let rise for 1.5-2hrs until increased in volume by half.
Shape dough into round and let rest covered with towel for 5mins, seam side down.
Transfer to bowl lined with lightly floured cloth or kitchen towel, refrigerate until nearly doubled, about 60-90mins.
Meanwhile preheat over with dutch oven to 500F, at least 30mins.
Carefully transfer dough seam side down into hot dutch oven, spray with water and score with parallel lines roughly 3inches apart, cover and cook for 15mins. Reduce oven to 450F and cook for another 15mins. Remove cover and cook for another 15-20mins until crust is dark brown.
Transfer to cooling rack and cool for at least 1 hour before slicing.

Notes:

Total time about 5-6hrs.
Can use stand mixer or mix by hand.

Step by Step

Amazing bread can be made with just a few ingredients. For this recipe it’s just flour, water, salt and yeast (and oil for greasing).

As for any baking recipe I find using a scale is the way to go.

Add the water to the flour and mix.

This mixture tastes awful. It really makes me appreciate how important salt can be, and offers a window into history and why salt was so valued (also for preserving meats) and caused the rise and fall of empires for thousands of years. If I had to eat bread with no salt I might go to war to get some salt.

Anyway, next step is to add salt and yeast and mix until the dough is smooth.

Cover and let rise for about an hour. After that take it out and form a ball - let rest for 5mins seam side down.

Place in towel lined bowl and in fridge for about an hour.

While it is in the fridge preheat the oven, with Dutch oven inside. Be careful when handling, something heavy and 500F is dangerous, especially when it looks the same when at room temperature and 500F - can you tell how hot it is?

For the record it was about 450F in that image. Next step is to carefully place the dough in seam side down, score the top, and spray with water. Scoring allows the expanding dough to expand somewhere, without it you might get oddly shaped bread or strange protrusions, plus it looks cool. The water is to make steam which helps create a hard crust, the Dutch oven lid helps keep the steam in contact with the dough rather than the sides of your actual oven.

After about 45mins of total cooking, 15mins at 500F with the lid on, 15mins at 450F with the lid on and 15mins at 450F without the lid you will get some nice looking bread.

This next part is probably the hardest, you need to let it cool on a wire rack for an hour. You just spent 4-5hours on and off making this bread, it smells amazing because it is fresh bread, and it looks amazing, but you have to wait a whole hour.

While cooling you can hear crackling noises, apparently that means it is really good bread. If you don’t believe me watch Ratatouille. Or make some and see.

The inside crumb shot:

You might be able to see one little area of undercooked dough, but otherwise it came out quite good. I’ve made this a few times and never been disappointed. I enjoyed it most as bread for grilled cheese - it makes a massive grilled cheese sandwich you can drench in butter and stack with cheese. It’s also pretty good as a side dish for pasta. Or as a snack by itself.