The ‘Sequel To SQL’ (The Proof Is In The Pudding!)

By Roland Schmidt, Senior Director of Business Development, Clustrix, Inc.

I am back again with a follow-up to my post on the Rackspace Developer blog from earlier this month wherein Clustrix drew the following conclusion:

“For highly-concurrent transactional workloads and real-time analytics Rackspace’s new Performance 2 flavor class Cloud Servers provide great performance, superior price/performance and therefore excellent value. Clustrix recommends them without reservation for running the ClustrixDB scale-out SQL database in a cloud.”

After extensive testing earlier this year and learning what we did about Rackspace’s new “Perf2” Cloud Servers, we decided it was high time to put our own recommendations to the test. So we approached a long-time common Clustrix-Rackspace customer, Agoge, Inc. about consolidation and migration of their production social network gaming environment. They seemed like a perfect candidate to benefit from now being able to run ClustrixDB in the Rackspace Cloud instead of running it as a DBaaS from a ClustrixDB Appliance, co-located in Rackspace’s ORD datacenter.

The Genesis of the Problem (the times they are a changin’)

Agoge ran everything but their database in the Rackspace Cloud. This was historical dating back several years. Prior to moving to ClustrixDB, and due to their high DB performance requirements, Agoge had co-located their own custom database server in ORD. But by mid-2012 it was running out of steam and they didn’t really want to buy, build or deal with another custom DB server. They came to Clustrix in order to get a scale-out SQL solution that was fully compatible with their MySQL code base, and could grow easily and seamlessly over time. Which it did just fine. But inefficiencies and cost issues surfaced as their business grew.

The Impact of the Problem (waste not, want not)

Because their App servers were in their own Rackspace Cloud account and the ClustrixDB Appliance they used as a DBaaS was co-located in our Managed Hosting account, significant growth on their site was creating much more network traffic between our two separate accounts. This, coupled with two firewall transitions for every database query on and off of the common network in ORD, was beginning to create excessive and costly network traffic and unpredictable bandwidth costs for Agoge based on unknown future traffic to their website.

The Simpler and More Cost-effective Solution (the Perf2 proof IS in the pudding)

By retargeting ClustrixDB from our co-located appliance to a powerful three-node cluster of new 30GB/8vCPU Perf2 Cloud Servers, all elements of Agoge’s system are now combined into a single cloud domain – in Agoge’s own account and behind their own firewall. They now have a simpler, more efficient and cost-effective service. And, they have preserved the ability to easily and seamlessly expand their database on demand, as ClustrixDB provides the same features and scale-out performance for Agoge in the Rackspace Cloud as it did for them using it as an appliance. But now it’s even quicker and easier using Perf2 Cloud Servers, as new nodes can be provisioned and added to their ClustrixDB cluster in just minutes.

Summary (creating the “win-win-win”)

Combined with eliminating the unnecessary network traffic and associated costs, Agoge’s new solution also lessens congestion on the ORD datacenter networks. This was an attractive win-win-win scenario for all three parties with newfound efficiencies enabled by Rackspace’s Performance 2 Cloud Servers.

To learn more about ClustrixDB visit our Rackspace Marketplace page and click on the Resources tab there to hear about the ClustrixDB architecture and How ClustrixDB works.

This is a guest post written and contributed by Roland Schmidt, Senior Director of Business Development at Clustrix, Inc., a Rackspace Marketplace partner. Clustrix provides ClustrixDB, the industry’s first scale-out SQL database engineered for the cloud, which is uniquely and ideally suited to handle massive volumes of ACID-compliant transactional workloads while concurrently running real-time analytics on the same operational data.


Please enter your comment!
Please enter your name here