Archives for January 2015

Direct-sales food distributor Epicure deployed a new server cluster to keep its websites from being overwhelmed — and saved money on SQL Server and storage in the process.

Traffic on the two primary websites operated by Canadian direct-sales food distributor Epicure went up substantially at the end of each month — so much that the spikes often made the sites unusable. Epicure tried working with the third-party vendor that hosts the sites to resolve the problem, but saw no improvement. And the North Saanich, British Columbia company’s business model is dependent on the two sites. It uses one website to market itself and its selection of food and cookware products; the other site is used by the independent “consultants” who sell the Epicure products at tasting parties to place orders and communicate with the company.

Russell Born, senior network infrastructure administrator at Epicure, looked for a way to keep the sites running smoothly through the end-of-month rush — and, while doing a Google search last year, he found SIOS Technology Corp. and its DataKeeper data replication software andSANLess clustering tools. The SIOS offerings enable users looking to guard against processing downtime to create Windows or Linux server clusters that utilize local storage only, avoiding the need to deploy a storage area network (SAN).

“Before that, I thought we required a SAN,” Born said. SIOS also supports SAN-based clusters, but Epicure would have faced some difficulties if [it] went that route,” he added. According to Born, a SAN leads to higher costs and more setup effort; it also creates a “single point of failure,” since the entire operation becomes dependent on it to function. To combat that, Born could have deployed multiple SANs, but he said doing so would have become prohibitively expensive.

No backup in case of outage

Epicure had another good reason to want to bolster its websites, which use SQL Server to process and fulfill orders and track the work of its consultants. Shortly before Born learned about SIOS, Epicure’s data center suffered an outage that lasted several hours. “We didn’t have redundant Web servers, so there was no failover externally,” he said.The company had two servers, one for each site, and both were reaching their resource limit, which contributed to the end-of-month performance problems. In addition to SIOS, Born turned to cloud computing vendorAmazon Web Services (AWS) to help relieve the pressure. The new setup for Epicure includes a primary Web server running on-premises and a cloud-based failover system, and the SIOS software replicating data between them. Concurrently, Epicure is upgrading from SQL Server 2008 to SQL Server 2012, and from Windows 2008 to Windows 2012.Jerry Melnick, COO of San Mateo, Calif.-based SIOS, said one of the primary motivators for the deployment was Epicure’s concerns about disaster protection and recovery. “There is so much depending on those sites,” he said. “Users are ordering off the sites. If they go down, it could take weeks to get [them] back up.”One method for ensuring high availability is to deploy servers in a second location so that, if the primary data center goes down, you can failover to the backup one. But Melnick said that has a high startup cost, so he recommended companies like Epicure tap into a cloud provider, such asAWS, for disaster recovery. SIOS’ job, according to Melnick, is to be the linchpin of the SANLess cluster.

Read more on TechTarget.com