Due to some major changes around here, we are winding down much of the activity here on the InfiBase blog. However, the State of the Cloud series of market research posts is scheduled to continue over at its new home, my cloud computing blog Jack of all Clouds.
I’ve just posted freshly-crunched numbers for September: State of the Cloud September 2009.
Hope to see you there!
Last month we launched our series of reports on usage of cloud computing infrastructures. Today we’ll follow up to see what has changed in the month that passed. We’re also delighted to expand our coverage into additional providers, providing a wider perspective of the cloud’s adoption.
To see our methodology and data sets, please refer to the first post in the series.
Of the 500k sites analyzed, these are the results for the five cloud providers we currently track:
The big surprise here is Rackspace, which seems to be just a step behind Amazon. We should recall that Rackspace’s Cloud Servers (Cloud Sites was not included) is actually Slicehost, an established VPS provider Rackspace acquired in 2008. SliceHost’s transition into a true cloud provider is still in process. For example, we cannot tell how many of the customers we’re seeing are paying per hour as opposed to paying monthly as they could on any other VPS. To highlight this, imagine GoDaddy rebranded their VPS/hosting services as “cloud” – they’d blow the others right out of the water (GoDaddy comprises 6.2% of our sample – almost 20 times the size of Amazon EC2!).
We were curious to see the impact of Google AppEngine, despite the fact that it is a PaaS not IaaS cloud. Despite all the hype, AppEngine has a mere 78 appearances in the top 500k sites.
As promised, we’d like to compare Amazon EC2’s standings since last month.
Overall, Amazon EC2 grew by a whopping 9% since last month. This is equivalent to 181% annual growth – truly incredible. Drilling down into the groups, we can see that it’s the first group and the last two groups that are growing fastest. The mid-range seems to be slower at adopting EC2.
Looking deeper into our data, we found 178 sites that migrated into EC2 this month but also 50 that left EC2. Newcomers to EC2 include GossipGirls.com, tech review site WebUpon and Central Connecticut State University. Departees from EC2 include National-Louis University as well as microblogging site Identi.ca (which, we found, switched over to EC2’s competitor Joyent).
As part of our tinkering in the cloud, we decided to run some benchmarks to figure out how well MySQL really performs on Amazon EC2. Some data has been posted here and there but no comprehensive results really gave us the answers we needed. So – we prepared a complex set of automatic benchmarks, ran them many times under various scenarios and analyzed the output. We’d like to share with you our findings (so that you won’t need to work hard like we did…).
In this post and the following ones we will provide the highlights of our benchmarks of MySQL on EC2. Please feel free to join the conversation, ask questions and add your own findings!
In this first post, we’ll share our methodology – and surprising results that not all instances are equal!
Our primary questions were:
We ran the following benchmarks using various configurations:
Additional notes:
In case you are not familiar with EC2’s instance types, here’s a short summary:
Now, to the results!
It’s a virtual world. Amazon EC2 sells compute units. So you might expect that all units are the same. Well, they are not!
You don’t know in advance which processor your virtual instance will be running. Even for a specific instance type, Amazon use both Intel and AMD processors, of varying types and clock rates. Why is that important? Well, better CPUs give better performance. We found consistently that the Intel-based instances gave better performance than the AMD-based ones – on the same instance type (i.e., same price!)
Intel Xeon 2.66GHz CPUs performed 30% better than AMD Opteron 270 CPUs using DBT2:
The conclusion: in the EC2 “roulette”, you should hope to get an Intel Xeon!
Below you can see the distribution of CPU vendors per instance type. It is interesting to notice that high-CPU instances always got Intels and small instances almost always got AMDs.
Overall we ran into the following CPU types on EC2:
So we’ve learned that results vary according to CPU type. Unfortunately, EC2 has several such inconsistencies. We found that the exact same benchmark on the exact same instance type and processor type can still get different results. Most results will be similar but once in a while you might see a significant difference. I/O to the ephemeral (local) disk and to EBS is also inconsistent and is affected by various I/O parameters, network performance more. I realize that the fact that it’s a virtual machine can cause these side effects, for example your physical server is hosting instances that run heavy processes (such as MySQL benchmarks!).
Now that we understand the basic ingredients of EC2 and the consistency (or lack thereof) of the results, we can continue to research additional aspects of MySQL’s performance on EC2. In upcoming posts, we’ll share our results for various instance types, MySQL versions and configurations and much more.
We’ve been talking recently to quite a few companies who decided to take the plunge and move their services into the cloud. Some were just starting while others had plenty of mileage under their belts. A major question we posed to these companies was, why did you switch to the cloud?
Here’s what we found.
The most frequent reason cited was that the cloud wins in cost. This isn’t fantasy, it’s reality: these companies significantly slashed their costs by switching. One CTO proudly mentioned his board of directors’ enthusiasm upon seeing the the new bottom lines. Overall, there was agreement that despite the initial price tag – some work required to switch – it’s totally worth it.
Scale up, scale down, on-demand, the sky’s the (theoretical) limit, ’nuff said.
The process of developing, debugging and testing is not fully addressed in a classic deployment environment. It gets easier by an order of magnitude in the cloud. Staging environments, parallel testing and ad hoc farms can be created and destroyed at the developers’ pleasure.
Although this usually wasn’t the primary motive for switching to the cloud (as opposed to the reasons above) it was a bonus many chose to highlight.
The cost of experimentation is very low in the cloud. This is making it easy for companies to test drive new concepts without making a large up-front commitment. If it sticks, the transition into a fully-fledged application is smooth.
Notable absentees in the motivation to switch to the cloud were the new but proprietary services offered by several providers. There appears to be a notable degree of resistance to vendor lock-in, which is the Achilles’ heal of services such as AppEngine, SimpleDB and SQS alike. This is perhaps the secret to Amazon’s success in the cloud – its strongly decoupled offering allows for a pick-and-choose approach, without major paradigm shifts – evolution not revolution. Companies clearly prefer standard tools based on tried and tested technologies.
With the explosion of cloud computing infrastructures, it remains unclear how many organizations and businesses are actually taking advantage of this new platform. Here at InfiBase we’ve been asking (as well as asked) that question regularly. The large providers aren’t going out of their way to satisfy the thirst of industry analysts and market researchers alike – so we decided to take a more practical approach and dig up the numbers for ourselves.
The report which follows has been collected and analyzed using proprietary tools based on public data. This post is a first in a series of regular posts where we will share our findings with the cloud computing community. Don’t forget to subscribe to our feed so that you don’t miss future updates!
The input dataset was QuantCast’s top site index. We took the top 500,000 sites listed and ran them through our scanning tools to build an index of the websites which are hosted on Amazon EC2. So, how many of the world’s top websites are placing their gateway to the world and, in many cases, their entire business – in the hands of Amazon’s cloud?
It’s easy to see that the larger websites are leading the pack and moving to Amazon first. The 0-100k group contains over twice as many websites on Amazon EC2 than the 400k-500k group. This trend is further emphasized if we drill down inside the first group. Our data shows 168 sites on Amazon with ranks between 0-25k, compared to 86 with ranks between 75k-100k. It would appear that usage of EC2 increases exponentially with rank.
This result makes a lot of sense as it’s the larger sites that are likely to benefit from what cloud computing has to offer – elasticity, the economy of pay-per-use and simply sheer scale.
In next month’s report, we will include a comparative analysis that will show how use of Amazon’s compute cloud is growing. We can already hint, based on some internal numbers we have collected in the past, that this growth is dramatic.
Our report does not purport to cover all cloud users nor does it encompass all users of Amazon EC2. Back-end and back-office systems, research servers and domains whose primary website (www.domain.com) are not on EC2 – are not covered. Additionally, our decision to focus on Amazon means sites hosted on other cloud providers are currently not included.
Nonetheless, these statistics are providing us us a fascinating glimpse into adoption of the cloud. We hope you find them as interesting as we did. Make sure to tune in next month to see the low-down on Amazon EC2 usage.
InfiBase is an early stage startup working on a revolutionary solution for the cloud. Our blog will serve to share tips, tricks and commentary on the world of cloud computing. As we are in stealth mode so we’ll do our best to refrain from blogging about the solution we’re building, although as you can imagine we’ll probably spill the beans here at some point or another!
Enjoy your stay and don’t forget to subscribe!
Britannia rules the clouds?
Well, we promised cloud-related posts, didn’t we? Let’s kick off with some comic relief before we get down to serious business.
Britons were excited a few weeks ago to find that the ol’ Empire had harnessed the power of clouds (as in sky) to paint a picture of the British Isles somewhere over the English countryside. Britannia Rules the Skies, the reports said.
(It’s actually the little sister Ireland which might rule the clouds (as in computing), due to the favorable tax laws which have attracted a lot of tech firms over the past couple of decades.)