Achieving the eCommerce Holy Grail – Outages with no Downtime
In the midst of penning this eCommerce blog series, we got this amazing note from Mouser Electronics, one of our eCommerce customers:
"A few days ago we sustained a sudden hardware failure which caused a hard shut down of our primary SQL Server node without warning. There were over 1,000 active sessions on the cluster at 4,000 requests per second. ScaleArc instantly load balanced the reads to the remaining servers and buffered the write requests while the SQL cluster passed Primary to another node. All kinds of alerts were triggered, but by the time the response team had rushed to their workstations, the ScaleArc/SQL cluster had already handled the situation. The recovery was fully automatic – there was no outage, no dropped sessions, and no tangible customer impact. Impressive." (emphasis added)
In sharing this experience, Mark Price, Mouser's director of Internet operations and architecture, captured the value of failover without any application impact - and therefore, without any customer impact.
Our eCommerce customers are, understandably, the most sensitive about downtime, since most suffer significant revenue losses as a result of even a minute of downtime. We’re very proud that so many eCommerce customers are benefitting from ScaleArc’s auto failover capabilities.
The eCommerce execs we've been talking to highlight two other sources of angst around uptime:
- Extending zero downtime to maintenance. Enabling to ensure online payments M-GO for its digital movie streaming business is crucial to the bottom line, but as with most companies, it suffers more from planned downtime for maintenance. ScaleArc’s ability to deliver zero-downtime maintenance has been a huge win at M-GO. “With ScaleArc, we can perform all of our maintenance tasks – even the ones that result in table locking – all in the middle of the day. We can redirect traffic away from our master nodes, and using ScaleArc’s failover capabilities, the application and our business doesn’t even see a blip,” says Vahram Sukas, M-GO’s vice president of live operations.
- Maintaining uptime while scaling new properties. Another eCommerce VP noted the challenges of maintaining both availability and performance as parent sites buy properties in adjacent markets. The infrastructure isn’t scaled yet, and the acquiring site instantly brings a huge new population to the new community. Enduring those growing pains and building consistent infrastructure and processes across them takes time, and the company is mindful of maintaining a good user experience across the growing family of product brands.
Learn more about migrating to a zero-downtime environment, and in our next post, we’ll discuss the limits of load testing and using the cloud to support eCommerce.
comments powered by Disqus