From the recently concluded Velocity Conference, all about Web Performance, theCube co-hosts Dave Vellante and John Furrier talked with Srini Srinivasan, VP of Engineering at Aerospike.

“People really want to see the latest and greatest performance and they do not want to wait, and that powers up performance,” says Srinivasan. Spped is very important for the business value. “Downtime is not an option at all.” Slowing down with just 20 percent ends up in losing a lot of users.

Seconds and the difference they make are real, but not in an absolute sense, rather in a relative sense, explains Srinivasan. Competition is rough, and so the servers with faster response times will gain customers.

Speed, Scale, and Reliability

Not too many people worry about vertical scale. “You want to squeeze the best performance out of every piece of hardware,” and that contributes to speed, Srinivasan says. Scale is putting together multiple such high-performance nodes to form a cluster. That allows you to enable people to add capacity on-the-fly. The server keeps running while you do these things. The self management of the software keeps the servers running at all times, with the proper SLAs in place.

The speed helps because there’s all this extra capacity in the system, and it enables you to handle unforeseen operations. When you have a lot of data in the system and you’re doing reads and writes, that goes on all the time. Sometimes a bad thing happens, such a node dying, but you still want the user avoiding a lengthy three-second response time. The extra capacity of the system allows you to tie it over, to re-balance the cluster data.

• Managing complexity

One of the fundamental decisions when founding this company was “keeping the architecture extremely simple,” says Srinivasan. Every single one of our nodes is self-managing, but it’s also the same with every other node. No hierarchy  The hard part is in the software, where the nodes have to deal with issues.

See Srinivasan’s entire segment below.