Database technologies, like MemSQL, are highly complex pieces of software that require extensive testing to ensure they meet stringent enterprise requirements. For example, MemSQL runs on anywhere from 270 and 400 nodes at Zynga, a social game services with over 240 million monthly active users. MemSQL must test its database across hundreds of machines to ensure a high level of quality for an environment of this size. For each line of code, the MemSQL development team utilizes thousands of automated test runs. MemSQL also leverages the AWS API compatibility to burst heavy testing workloads into the public cloud and bring them back in-house when extra capacity is no longer needed.
MemSQL needed an on demand dev/test environment that could withstand heavy testing with no latency for its QA initiatives.
We run thousands of tests each month. Our product depends on the quality of our testing and how fast we can perform our test runs —poor QA would kill our reputation in the industry.Nikita Shamgunov, co-founder and CTO of MemSQL
The Business Benefits
Key benefits MemSQL has achieved as a result of deploying Eucalyptus’ private and hybrid solution for development and testing include the ability to:
- Utilize in-house hardware to create an on-premise cloud solution with existing resources
- Easily run applications and tools on-premise that have been developed for AWS
- Deploy and manage mission-critical development and test resources in a scalable, self-service and on-demand fashion
- Source code to resolve glitches and troubleshoot due to open source technology and community
- Provision servers on demand to meet specific needs and allow infinite scalability
- Provide run times at a rate of three times faster than testing in a public cloud, with no latency
- Create a test environment to instantly spin up instances and terminate after test cycle is complete
- Migrate virtual servers between Eucalyptus and AWS
Eucalyptus provides the ideal testing environment and is a key part of our software innovation process. With Eucalyptus, we are able to quickly bring products to production and realize an immediate return on investment. We plan to continue to grow our Eucalyptus implementation by at least 20 percent in the next year.Nikita Shamgunov, MemSQL
We had a lot of underutilized hardware sitting around and realized we could save $250,000 a year by building our own on-premise solution with this existing hardware," said Shamgunov. "We chose Eucalyptus because of its ease of use and AWS API compatibility. We had Eucalyptus up and running in a weekend—it just worked!"
The Eucalyptus Cloud platform is open source software for building AWS- compatible private and hybrid clouds. Eucalyptus leverages MemSQL's existing hardware and networking capabilities to satisfy the company's rigorous development and testing demands.
Using Eucalyptus, MemSQL has created its own testing cluster in-house to ensure a high level of elasticity and scalability with little to no latency.
"Prior to implementing Eucalyptus, we had to manually configure a set of nodes for each distributed test, which was a very slow and painful process for the engineering team," said Shamgunov. "Implementing a private cloud has completely transformed this process."
AWS API compatibility was a key factor in MemSQL's selection of Eucalyptus. This fidelity enables MemSQL to reuse existing AWS-compatible tools in its private cloud, accelerate testing cycles, and maintaining IT governance and control. MemSQL currently runs more than 400 virtual machines between its public and private clouds.
MemSQL offers a distributed in- memory database that provides real-time analytics on big data for customers including Zynga, Morgan Stanley, Hitachi and VeriSign. The technology empowers organizations to make data-driven decisions, better engage customers and discover competitive advantages. MemSQL was built from the ground up for modern hardware to leverage dozens of cores per machine and terabytes of memory. The company’s real-time analytics platform is deployed in production environments across thousands of nodes in the most high-velocity big data environments in the world.