hylife

Some pigs are known for their ability to sniff out precious truffles in the forest, but for Manitoba’s HyLife, a global producer and distributor of pork products, finding the answer as to why virtualization was slowing down its ERP systems took a bit of digging.

Outgoing senior systems administrator Marcel Broesky grew up in rural Manitoba, a province that has relatively small IT community, and when he joined HyLife, virtualization was very much an afterthought. He spent the first year and a half building out the company’s VMware infrastructure to provide redundancy for its ERP, but from day one, he said, it was hit by performance problems.

HyLife, which has about 1,700 employees and 500 discrete email addresses and desktops, has two ERP systems – a custom one specific to hog production and Microsoft Dynamics – to pull various forms of data from disparate systems. This data includes ham scale, loin scale, bacteria levels, plant activity, and more. With all of its core applications and databases virtualized, HyLife had a more versatile and cost-effective infrastructure, but it was causing queries to take longer expect – up to 2.5 minutes on average for each report.

“We couldn’t figure out the problem,” said Broesky. The quest for a solution led him to VMworld, where he spoke to as many people on the tradeshow floor as possible where he encountered PernixData, who informed him he had a data latency problem.

HyLife ended up using PernixData FVP software to solve its database performance challenges by installing FVP inside its vSphere kernel to support clustering of several Fusion I/O cards across various Cisco UCS blades, said Broesky. This created a low latency acceleration tier for storage reads and writes without having to change the company’s existing Cisco and IBM storage arrays. “Immediately the benefit was obvious,” he said. “It improved our systems so much that we continued to grow our ERP system by injecting more data into it.”

The IBM flash storage system employed by HyLife is the same one used by a popular dating site Plenty of Fish, noted Broesky, which sees in excess of 150,000 new users a month. “We may not be a popular online dating site, but we have a lot of unique users a month.” Those users come in the form of about 130,000 new hogs, and the data surrounding all of that pork is queried a lot.

While Broesky concedes HyLife’s software could be more efficient – “it’s not away best practice to throw hardware at software issues” – the company’s developers are focusing their attention on other efficiencies.

In the meantime, HyLife has been able to address potential downtime through virtualization and redundancy across two sites, which would cost the company $2,000 per minute, Broesky said. If you have a major failure, you’re looking at two or three hours. That adds up pretty quick.”

A massive query used to take hours, but that time has massively shrunk, he said. “Data is constantly getting flushed through.”



Related Download
A Simple Powerful and Practical Analytics Solution For A New Generation of Users Sponsor: IBM
A Simple Powerful and Practical Analytics Solution For A New Generation of Users
Download this report for an in-depth look at the capabilities and features of IBM Cognos Analytics.
Register Now