http://www.thebitsource.com/2010/01/11/rackspace-cloud-servers-versus-amazon-ec2-performance-analysis/

I'll say that I've had this conversation dozens of times with various people and it's nice to see someone took the time to really do a good study. They really went all out to test under various conditions, times of day, different servers on the same platform, and compare apples to apples about as much as is possible. One of the sage things in the report is this gem, "Typically it is best to test specific applications to get a true measure of performance..." That is so very, very true for many reasons. I wish they had time to do the test on the Joyent platform as well. It's similar in it's CPU burst-ability but under the hood it's quite different. Info I provided about this was the subject of a couple of blog posts a while back here. For example, their platform uses 8 cores whereas Rackspacecloud uses 4 cores. Amazon, as far as I know, doesn't allow you to burst at all your CPU allocation is a hard cap which is quite different from RackspaceCloud, Slicehost, or Joyent.

Lastly, people forget and I keep saying ad nauseum, how you write the app means everything as to how well it will perform on the cloud. Generally speaking, adopting many of the ideas and processes around development of distributed applications is key.  But, this is also a very foreign concept to the bulk of developers I've worked with over the years.  This fact, how you write the application matters, is true pretty much no matter what cloud you deploy on and intimately tied to how well you automate certain operations tasks as well. Anyway, I know I've been a quite blogger lately. I simply have too much work to do to blog often with my startup. But, hopefully I can get back into the groove sometime soon.

Cheers!  -Kent

Very interesting.  If it was pure spot pricing then, in the event capacity became scarce in the context of demand then more money would flow to AWS as prices rise to help expand capacity.

At the moment a us-east-1, m1.small is $0.026 / hr.  This is still more that the $0.015 cost of a similar device in the Rackspace cloud.  So, to do a small video encoding job I'd still use Rackspace.  But, that can change pretty fast perhaps.

The one thing this spot pricing doesn't take into consideration is actual CPU resources available relative to price.

There was an analysis, I'm searching for the link will add in the comments if I find it, recently comparing different cloud servcies, their cost, and the amount of vCPU you get.  For example, if I say that the us-east-1 m1.small is $0.026 and the Rackspace equivalent is $0.015.  But, I get 4 vCPU's at Rackspace and only 1 vCPU at Amazon, then in reality Amazon is still far more expensive than Rackspace above and beyond the spot pricing vs. fixed pricing if you measure based on CPU Availability per dollar.

This is not a new problem.  It's one that is often quite overlooked when comparing the value of one cloud vs. another cloud.

The list of links from this podcast episode was particularly intriguing for me this week.

Of particular interest to me is:

TorqueBox - JRuby backed Rails application platform (and more) build on JBoss AS.  Very intriguing and I'll be experimenting right away with this for an application I've just pushed out into production.

ShardTheLove - An Active Record horizontal sharding solution with build in support for migrations, testing, and more.  I will also be evaluating this for inclusion into a new application I am just launching.

Jammit - A static asset packing solution for Rails applications. Finding a good solution for this can sometimes be challenging.  However, doing it in any modern web application is pretty much mandatory.  I look forward to testing this library.

Great stuff and worth a look if your pumping out Rails applications that you want to be scalable on-demand.

The Overall Monitoring Architecture

There are areas of monitoring that I usually like to pay close attention to with a live web application.

1. Process Monitoring - This makes sure things are running and stay running within certain tolerances. Examples are God, Monit, SMF.  Your choice will depending on your operating system and preferences with scripting languages.
2. Resource Monitoring - This is fine grained CPU, Memory, Disk Space, Disk IO, Networking, application server threads, and much more. Examples are Nagios, Ganglia, and Munin. Choosing correctly depends on your specific situation.  There is a worth newcomer on the block called Reconnoiter that also looks very promising.
3. UpTime Monitoring - This is the only monitor people usually do if they do any at all. This should be a disinterested 3rd party to provide accountability and what I call a 3rd party eye in the sky should any dispute about uptime arise.  I like pingdom and there are even free services as well.  I’ve also been using CloudKick in some situations for this purpose as well.

Those three above are from a post I wrote some time ago.  Today, I’m adding a 4^th item to that list because it has finally become easy enough and reasonably affordable to add now that there is an affordable choice:

4. Synthetic Transaction Monitors – These actually perform tests of processes a user might go through in your application and report back any anomalies if they occur along with an error report, screen shot, and other data as appropriate.  I’ve been using a tool called BrowserMob and Selenium IDE for this.  You create scripts w/ Selenium, upload them to browswermob and then setup a monitor script.  That’s a simplified overview of course but it’s really quite effective and relatively affordable compared to historical solutions for synthetic transaction monitoring.  Historically it was prohibitively expensive to do synthetic transaction monitors.

The Monitoring Tools I use

What follows are some of my personal current favorites to meet the above goals.

Munin > http://munin.projects.linpro.no/

One of the things on my list for a while to get done is enable munin across your systems.  I use it a lot successfully.  You can see a demo here:

Monit > http://mmonit.com/monit/

Pingdom > http://www.pingdom.com

Things I’m testing and have high hopes for are Reconnoiter, BrowserMob monitoring, CloudKick

Enterprise Cloud Computing Startup + Sr. Systems Administrator (sausalito)
http://sfbay.craigslist.org/nby/sad/1366310094.html

Cheers!b

Kent

My comments:

Excellent write up. Thank you. This is almost exactly what I advocate day after day regarding planning for scalability. I like that you’d put a bit of a framework around it.

Personally, I treat scalability concerns the same as any other “feature” in a project in a Agile Development context. It’s brought up and discussed briefly in scrums, possibly handled in a follow up meeting, and then either put on the backlog to be prioritized w/ everything else or worked on next if necessary.

It’s often very, very difficult to get some teams to think pro-actively about scalability AND agree to table it for later. I think this is because sometimes when you have these discussions people realize that they need to “fix” something and can’t really help themselves.

Of course, these days with cloud computing gaining so much traction and resources being available on-demand more than ever at a moments notice it’s getting easier. But, if you don’t architect for scalability at the outset you may find when you get the implementation and deployment phases that it’s impossible to do what needs to be done. But, that’s a whole other topic I suppose.

Cheers!

You can see the results of my efforts here:

http://www.legalcloud.net/buzz.php

As for how I did this, it wasn't terribly difficult and I'm quite happy with the results. I decided to use a Rackspace Ubuntu 9.04 cloud server of the 1G size. You could use any server you want, this was just what I chose to use. Using your manage.rackspacecloud.com account you need to do the following:

• Start an Ubuntu 9.04 server
• Log in to server as root
• apt-get Update
• apt-get install ffmpeg

Transfer your WMV file from where ever you put it to your new server.  I uploaded mine to Cloud Files then just downloaded them from there to encode.

The following command can be used to convert a GoToMeeting Webinar (not the proprietary format) to an FLV.

ffmpeg -i "tdp.wmv" -sameq -ar 22050 -ab 96000 -deinterlace -nr 500 -s 720x576 -aspect 4:3 -r 20 -g 500 -me_range 20 -b 270k -deinterlace -f flv -y "tdp.flv" 2>tdp.txt

This command converted the 26MB file in 2-3 minutes for me which I was happy with. I tried this with local desktop tools on a mac and it took well over 1 hour (and wasn't mostly free).

The whole process takes under an hour so you can do this for approximately $0.06 which seems good to me since it's only something I do occassionally.

Note: You cannot use this command to convert the GoToMeeting proprietary format. I haven't found anything that would yet but I hope to. So, make sure you don't choose the compressed proprietary format for recording your meeting!

Once the conversions were done I uploaded the resulting FLV files to Rackspace Cloud Files and enabled the CDN feature for them. Then, I integrated them with our Flowplayer installation on my web site. I'm quite happy with the result and the Flowplayer team deserves massive amounts of credit for their very cool and usable software that building upon jquery.

I'm not an ffmpeg expert but this did produce nice results for me and it's what I came up with after some trial an error. If anyone has any suggestions to make it even better then please let me know.

This same process would work just fine with any cloud provider of your choice really so if you use GoGrid or AWS the same steps should apply. I can't speak for other Linux distros at the moment but I'm sure you'd have good luck there too.

Cheers,
Kent

Service Models:

LegalCloud is an IaaS model cloud. We are specifically delivering data center infrastructure to law firms on a globally.

In the not so distant future there will be PaaS and even possibly SaaS opportunities associated with and deployed by LegalCloud and it's partners.  There are many possibilities.

Deployment Model:

LegalCloud is a Hybrid Cloud composed of both community and private types.

Our Community is Law Firms and only law firms. This focus allows us to uniquely and completely address the needs of our clients.

As a Hybrid cloud, we provide both on-premise and off-premise services for our customers that bridge the gap between their own facilities and the cloud facilities we manage as is appropriate and necessary.

Essential Characteristics

The LegalCloud console (1.1), which is in Alpha at the time of this writing, is the tool that our clients use to 1.0 LegalCloud Console Default Viewself-provision servers in any of our globally distributed data centers. For the first time publically I’ve included a couple of small screen shots from our staging environment. The things they provision are networking, compute, storage, and a few other related things. The storage components in particular are interesting because they can be further dynamically provisioned and grown (or shrunk) on-demand.

The resources that clients provision via our console are from pools of resources. In our case they are not truly location independent as we must provide a certain amount of auditability. But, they are deployable in various geographies.

Rapid elasticity is primarily a function of programmatic interaction w/ API based controls. We will not have API access for our first release. But, we most certainly will layer it in over time. Now, which one to pick?

Our console in association with something we call a pod manager is essentially a part of a distributed monitoring tool that allows our clients to keep an eye on what’s going on for key metrics in their pod.1.1 LegalCloud Global Servers View

LegalCloud has an currently uncommon “broad network access” model. It’s production environments are only available to clients via secure VPN technologies or private lines (point to point or MPLS). We do not allow general access via the internet at large. Period. Within legal cloud all clients, while they do share some infrastructure, they do not co-mingle their data/networks.

That wraps up my comparison of how LegalCloud can be fitted to the NIST cloud computing model.

What’s next?

What is missing from the NIST model today, if it belongs there at all, are the security aspects. I have seen what is likely to be important and solid work going on around an initiative called A6. It discusses Audit, Assertion, Assessment, and Assurance API. This is also now known as A6. There is a great amount of discussion going on in this arena and I’m looking forward to analyzing LegalCloud relative to the A6 API as it matures.

So,  as soon as possible, I will write about the other issues around security related concerns and some of the issues that matter to our clients around varous A6 stated concepts.

Here is an AR business card that mashes up with Twitter.

http://artimes.rouli.net/2009/07/cool-augmented-business-card-from-toxin.html

and more...

As soon as you can see these augmented reality images easier without a phone, webcam and a computer, or other unwieldy device that isn’t supremely geeky the world is going to change a great deal very fast.  Today those

http://www.vuzix.com/iwear/products_wrap920.html

devices still look kind of like this:

But, that's changing very fast.  Here is a contact lens that was recently "installed" on an unsuspecting little bunny rabbit.  It's a low resolution heads up display.

Then, of course, there is actually the concept of controlling things virtually and even in the physical world through a combination of AR and monitoring brain waves.

Neurosky is developing tools to translate feelings (brain waves) into actions.  What if you were to:

• Hook it to your car cruise control
• Get the will it blend guy to wire this up to his blender, will it mind blend?
• An augmented reality enabled version of the little engine that could where the kid actually participates in getting the train up the hill
• An alarm clock you CAN'T turn off until you are awake and "think" the damn thing off
• Obvious implications for additional controls for video/console games
• A mind controlled vibrator (so obvious)
• A lawn mower. Lawn Mower man 2.0.
• A vacuum cleaner
• Bank Lock boxes that require you to be calm to open them
• Drunk/Sleepy driving accident/construction prevention (what are a drunks brain waves like anyway?) - Car says your asleep and calls you a cab.

Depending on how difficult it is to learn to use this device of future revisions of it, it clearly has many possible uses. Why just calm? Why not also allow it to do things based on other states of mind like rage, excitement, happiness, sadness. That could be interesting!

As is often the case porn, marketing, and gaming will lead the way.  AR Strippers have arrived as well:

http://artimes.rouli.net/2009/08/ar-strippers-oh-my.html

Beyond Reality - AR Gaming is developing interesting AR games.

http://www.augmented-reality-games.com/

While the AR strippers are entertaining I suppose, the reason it is interesting is because if you imaging using a live feed in real time you see quickly that the future of video conferencing and phone calls in general could present a lot more of you than just your face and voice.  In the future, it'll feel to the people that see you in their spaces real and virtual that you are there.  Sometime well after that, it'll feel like you are there too.  But, that's much further out probably becuase integrating senses other than vision and audio is quite a bit more challenging at the moment.  We just don't understand the brain well enough yet.

Lastly, if you haven't picked up and read Rainbows End you might just want to as it could be a guide to how things are going to work in the not so distant future in several ways.

I wanted to do a quick post this morning to discuss this since it is almost entirely my focus these last few months.

Total Data Protection is the name of our Enterprise Class Hybrid Cloud Computing service that provides the ability for any Law Firm to provide business continuity for their enterprise compute workloads no matter where they are by leveraging our software stack and Private/Community Cloud deployments throughout the world.

In that definition of Total Data Protection I used some deployment model terms from the NIST definition of Cloud Computing; draft v14.  To review, those deployment models are:

Private cloud. The cloud infrastructure is operated solely for an organization. It may be managed by the organization or a third party and may exist on premise or off premise.

Community cloud. The cloud infrastructure is shared by several organizations and supports a specific community that has shared concerns (e.g., mission, security requirements, policy, and compliance considerations). It may be managed by the organizations or a third party and may exist on premise or off premise.

Public cloud. The cloud infrastructure is made available to the general public or a large industry group and is owned by an organization selling cloud services.

Hybrid cloud. The cloud infrastructure is a composition of two or more clouds (private, community, or public) that remain unique entities but are bound together by standardized or proprietary technology that enables data and application portability (e.g., cloud bursting).

LegalCloud.net is a true Hybrid Cloud as it is a combination of Private and Community and provides services both on-premise and off-premise.  It is shared by an organization, that organization is the aggregate of all Law Firms.  If you are not a law firm, you can't use LegalCloud.net.  Period.

We are working very hard to address all the common concerns about enterprise cloud computing.  We specifically address things like auditing, compliance, network security, data security, transparency, data location and the legal Issues surrounding it.

We have other products related to or complimentary to Total Data Protection on the way and in testing.  We'll be deploying our client facing console, a really cool distributed Rails (on the front) and Java (in part of the backend) application, in a very short few weeks to the first clients.  Clients will be able to deploy Total Data Protection, Active Servers, and Provision storage in our globally distributed data centers through this interface.  Our first release will not have a clent facing API unfortunately, but we're trying not to boil the ocean you know.  However, I have started working on this by studying the best of the available API's out there and expect to move forward on specification and early development stages soon.  Of course, the API will not be public, it'll only be available to members of our cloud commnity; law firms.  But, that is the point in our case.

When I started nScaled I never imagined I'd be building a cloud quite like this one.  But, it's exciting to be sure.  My blogging certainly has taken a hit but that's okay I suppose.  Over time I'll be able to blog more and more about the various things we've been doing.

Kent now returns to his usual daily program of coffee, phone calls, infrastructure, and sales calls... 

---Kent Langley, CTO, www.legalcloud.net by nScaled, Inc.

Authentication Request

curl -D - \
  -H "X-Auth-Key: youneedtoputyourownauthkeyhere" \
  -H "X-Auth-User: acctusr" \
  https://auth.api.rackspacecloud.com/v1.0

Authentication Response

HTTP/1.1 204 No Content

Date: Sat, 25 Jul 2009 15:44:07 GMT
Server: Apache/2.2.3 (Mosso Engineering)
X-Storage-Token: 8gc00ld-a77r-6548-eq58-5nb5hsrv9876
X-Storage-Url: https://storage.clouddrive.com/v1/MossoCloudFS_8gaserd-65q3-5dfas-8888-c66e9987r953
X-Auth-Token: aq9beanS-d75c-9135-ed78-4bs3chse9135
X-CDN-Management-Url: https://cdn.clouddrive.com/v1/MossoCloudFS_d96e4c99-85ej-2pol-7777-v33e9135e669
X-Server-Management-Url: https://servers.api.rackspacecloud.com/v1.0/928875
Content-Length: 0
Connection: close
Content-Type: application/octet-stream

These aren't real tokens so you can't use them yourself of course. You need to get your own, which is exactly the point. Once you have successfully received your own response then you can move on to actually using other API calls to do things with the API. Here is a cloud servers API example using curl again.

Listing Your Account Limits

curl -D - \
  -H "X-Auth-Token: aq9beanS-d75c-9135-ed78-4bs3chse9135" \
  https://servers.api.rackspacecloud.com/v1.0/928875/limits

That will output something that looks like this:

HTTP/1.1 200 OK
Server: Apache-Coyote/1.1
vary: Accept, Accept-Encoding
Set-Cookie: JSESSIONID=D6DBBE0BE05C4559BC71E6F851501575; Path=/v1.0
Cache-Control: s-maxage=1800
Last-Modified: Sat, 25 Jul 2009 15:51:26 GMT
Content-Type: application/json
Content-Length: 657
Date: Sat, 25 Jul 2009 15:51:26 GMT
X-Varnish: 1753447558
Age: 0
Via: 1.1 varnish
Connection: keep-alive

{"limits":{"absolute":{"maxTotalRAMSize":51200,"maxIPGroupMembers":25,"maxIPGroups":25},"rate":[{"value":25,"unit":"DAY","verb":"POST","remaining":25,"URI":"\/servers*","resetTime":1248537086,"regex":"^\/servers"},{"value":10,"unit":"MINUTE","verb":"POST","remaining":10,"URI":"*","resetTime":1248537086,"regex":".*"},{"value":600,"unit":"MINUTE","verb":"DELETE","remaining":600,"URI":"*","resetTime":1248537086,"regex":".*"},{"value":3,"unit":"MINUTE","verb":"GET","remaining":3,"URI":"*changes-since*","resetTime":1248537086,"regex":"changes-since"},{"value":10,"unit":"MINUTE","verb":"PUT","remaining":10,"URI":"*","resetTime":1248537086,"regex":".*"}]}}

Here is a very simple bash shell script you could use for the Authentication example above:

#!/bin/sh

curl -D - \
  -H "X-Auth-Key: 5115a48a9ca852bc266ea3e7bc8805e7" \
  -H "X-Auth-User: nscaled" \
  https://auth.api.rackspacecloud.com/v1.0

I know that there are at least a couple of companies with growing support via their cloud management platforms quickly adding Rackspace Cloud functionalities.  So, while I have to hack around with curl today I suspect that in short order it might get easier.

Cheer!

Kent

Launch a shiny new cloud server w/ the ubuntu 9.04 server template provided by Rackspace.

apt-get update
apt-get install libmysql++-dev make gcc+ g++

Go to a directory of your choosing

Download The Sphinx Binary (latest as of this writing)
http://sphinxsearch.com/downloads/sphinx-0.9.8.1.tar.gz

tar zvxf sphinx-0.9.8.1.tar.gz

cd sphinx-0.9.8.1

./configure

make

make install (you will need superuser rights for this last step)

The binaries will be install in /usr/local/bin

/usr/bin/install -c 'indexer' '/usr/local/bin/indexer'
/usr/bin/install -c 'searchd' '/usr/local/bin/searchd'
/usr/bin/install -c 'search' '/usr/local/bin/search'
/usr/bin/install -c 'spelldump' '/usr/local/bin/spelldump'

some config files will drop into /usr/local/etc

/usr/bin/install -c -m 644 'sphinx.conf.dist' '/usr/local/etc/sphinx.conf.dist'
/usr/bin/install -c -m 644 'sphinx-min.conf.dist' '/usr/local/etc/sphinx-min.conf.dist'

Enjoy!

What are you waiting for?  Check out the new release!

I've been doing a good bit of work with Rackspace cloud servers for the last few months for various projects.  Rackspace does let you create and properly apply DNS settings to a cloud server for the purpose of sending mail to clients.  This is nice in many cases as it allows you to avoid setting up an espensive server elsewhere or using a 3rd party relay service.

Be aware that there are any number of things I have left out or not discussed here. This isn't really meant to be a thorough how to article. This is just an overview showing something that is possible.

Here are the very high level steps and links to some of the pertinent resources you'll need.

1. Launch a Server from your manage.rackspacecloud.com portal
2. Install and Configure PostFix on that server.
3. Set your DNS Records
   1. SPF -Sender Policy Framework. You cannot skip this or your mail will not get delivered.
   2. Reverse DNS - This resides on the Rackspace side.  You cannot skip or mail will not be delievered.
   3. MX - This is done at your DNS server/provider.  Mandatory as well.
4. Check your firewall settings, there are firewall issues to consider
5. Send Mail happily from that server, or do a little more work and set it up as a relay server for your environment.  But, make sure you're not an open relay please!!

Random thoughts while I was typing this post:

• $0.015 per hour to start for this service! No per message pricing.
• If you start getting busy you can resize the server using the GUI resize tools
• If you're using some of the beta IP grouping features you can build an HA service easily
• Integration of cloud files for log storage should you wish to keep them
• Support is available and solid. They will help you if you need help. They mean it when they say they are fanatical about the support thing.
• If you want to get some monitoring on this (or other) boxes, use munin. Works great.

That is all for now.  Enjoy!