In The Cloud, Scale And Performance Are Imperative

By Korey Lee, CIO, SumAll

SumAll was built on the premise to help companies that are putting droves of data into the cloud on various platforms without know-how or engineering resources to get their data back out. We’ve created a real-time analytics platform for marketers, small and medium sized businesses (SMBs), and entrepreneurs. We strive to make their data easy to access and visually appealing without the help of an analyst or engineer so our customers can gain meaningful insights from connecting their data in one place.

Whether it’s data from Facebook, Google, PayPal, Stripe, Amazon or Twitter; we present it in a simple, tangible way for our users to digest it and make intelligent decisions based on it.

Since we formed in May 2011, we’ve crunched over 4 terabytes of data, and that’s growing at a rate of about 30 percent per month. We track more than $4 billion in revenue, 290 billion social actions and 190 billion sites for more than 150,000 businesses.

Like many startups, we started with our MongoDB stack on AWS. At the time, it was right for us. But as we started to grow, the costs required to scale skyrocketed. We quickly learned that AWS is great for small companies that are uncertain about how fast they’re going to grow; but once these small companies hit a growth spurt, AWS becomes increasingly expensive and difficult to optimize for performance.

In light of our non-traditional data growth, we started seeing performance issues early on, well under two years since our product launched. I/O instances were overloaded and we started to lose data – our customers had issues loading their existing data and pulling in new data. When your business is built on presenting data to your customers, losing that data can be catastrophic, not to mention compromising to our brand and market presence.

We moved our MongoDB stack to ObjectRocket by Rackspace. The persistent SSD storage gives us and our customers faster access to data and helps us better manage our growth. The purpose-built solution gives us a massive performance boost at a lower cost.

One way we measure performance is by import times. With AWS, imports could take days, and for some customers with longer data histories up to weeks. Once we cut over to ObjectRocket, those imports were whittled down to an average of 22 seconds, that’s magnitudes of hundreds of percents faster. At the same time, the volume of data streams, defined as each platform a customer adds, we could process with AWS topped out at about 1,000 streams per day. Now, we can hit 3,000 streams per day without breaking a sweat. If we had stuck with AWS for our MongoDB, we could not have continued this tremendous growth path.

For us, it really comes down to serving our customers. Our customers log into our tool on a daily basis. They use SumAll to run their businesses better. If we have blackouts or it appears to the end user that their data is flat lining, that kills trust. They trust us to deliver accurate data on reliable and real-time basis.

At the same time, our service and support objectives align with Rackspace’s Fanatical Support approach. We value it greatly, compared to a do-it-yourself environment. We vow to our customers to never outsource customer service. Each SumAll employee answers a certain number of support calls so we can stay in tune with what our customers experience – if we don’t have them, there would be no business. We set a strict goal for response times of less than 24 hours.

It’s that value-driven service and our ability to increase performance while reducing costs that has helped us scale and grow our business; and we’re confident that we’ll be able to continue that growth into the future with Rackspace and its team of specialists on our side.

This is a guest post written and submitted by Korey Lee, CIO of SumAll, a Rackspace customer. SumAll is a connected data tool that helps thousands of marketers and SMBs make better decisions.


Please enter your comment!
Please enter your name here