|
A global hedge fund implements PeerSync to efficiently replicate huge data sets – while maintaining optimum system performance.
Santa Fe Partners is global hedge fund that trades in the U.S., Europe, and Japan, and has an office in the Netherlands. They are, according to IT administrator Terry Rickard, “a small, lean, mean operation. We do with 7 or 8 people what larger organizations do with 15 or 20.”
The small IT group is entrusted with managing the huge volumes of data that sluice through the organization on a daily basis. Security is, quite naturally, a top priority, as is assured data availability (security concerns are such that Santa Fe doesn’t even have a public website). If they’re looking for a specific application or utility, they will typically write their own code. However, if they find a commercial piece of software that meets their stringent standards for security and efficiency, they will purchase it, even if it comes with a hefty price tag.
Recently, Santa Fe’s IT group was looking for a file synchronization tool to manage their huge, constantly changing data sets…a tool that could comfortably manage market and transactional data and allowed the company to transfer large amounted of data to remote disaster recovery servers. PeerSync Software allowed an annual cost savings of approximately $150K by reducing the need for extra bandwidth, regional disaster recovery servers and storage appliances.
The amount of data that needs to be downloaded and transferred daily exceeds 600 gigs. This required a dedicated T-1 line, as the process demanded 100% bandwidth. In addition to synching – in real time – data to and from their U.S. and overseas offices, Santa Fe backed it up to a remote server in a secure location in Atlanta.
The IT team sought a synchronization solution that was reliable, efficient, not bandwidth-intensive, and transparent to users. In order to satisfy these requirements, it needed to offer byte level replication, multithreading to handle simultaneous “jobs,” bandwidth throttling to allow for more control, and time out settings…to name the most critical features.
The Evaluation Process
Before joining Santa Fe Partners, Rickard worked for the state of New Mexico and was instrumental in developing a statewide data disaster plan. In that role, Rickard had experience with several pieces of replication software.
“I was familiar with the available tools – their strengths and weaknesses – when I helped in developing a disaster recovery plan for the state. State agencies require that you send requests for proposals to a minimum of three bidders; everything needs to be properly documented, so I became very familiar with the software on the market.”
Realizing they’d be better served reaching out to a software vendor than in writing it themselves, Terry and his team proceeded to test 10 applications, from the replication engine built into Microsoft’s System Center Essentials to the tool offered by Double Take
“MS System Center Essentials uses the replication that is built into the server, and replicates data across multiple points. It requires SQL, front and back end negotiations, and requires a lot of overhead. It uses distributed file management – you can point at a folder and distribute it to another network location. However, the overhead for making it work requires another server in your network, which alone can cost up to $5K just to handle these interactions.”
While Rickard indicated that cost was not a major factor, “no organization can justify what seemed to be an unnecessary expense.”
“Double Take, another tool we evaluated, cost three times as much as PeerSync, which at this point was another vendor we were evaluating. Second, it didn’t appear to have the flexibility and ease-of-use we were looking for. As an IT admin, I need to figure out how to use a piece of software in the first 3-5 minutes. It’s just not worth my time figuring out.”
Among the criteria the software needed to meet, true byte level replication was possibly the most important – the ability to point at a particular file or folder and replicate only the file you want. “Only found two that would do it to our 100% satisfaction. One of them was PeerSync.”
“Just to give you an idea,” continues Rickard, “previously, we had to allocate 100% bandwidth on a T-1 all day long. Using PeerSync on a T-1 requires 33% bandwidth. .
In this instance, and in others, PS did exactly what it said it would do. Hands down, from a features and performance standpoint, even a price standpoint, it was better – more efficient, less overhead, and less administrative cost on my part, which is a big deal. It was like going from two to four wheel drive.”
With erratic support the norm, Rickard was floored to find Peer Software’s responsive, professional, and thorough. “It was absolutely astounding: I call on the phone, I get somebody. A piece of software or hardware is only as good as the people you get. I’ve worked with Dell, CompUSA, Best Buy…you can buy a computer off the shelf, it’s only as good as the warranty. I don’t care how good your software is…I had to call all these companies for tech support…Peer Software provides far and above the best technical support. I used to write software, beta test…if I know it better than you, why did I call you in the first place.”
This would be a happy story if it ended there. But since implementing PeerSync for one set of tasks in mind, Rickard found it useful in a completely unforeseen way.
The IT group needed to convert a large amount of data from one version (4.0) to a new one (5.0), while making the updated data format available in real-time, without any interruption in business processes. Data needed to be moved from a production server to a staging server, then to a disaster recovery center in Atlanta. To provide a sense of the project’s scope, 300 gig of data would need to be copied manually if not for a tool that could do file level replication within seconds of a file being converted.
“We realized we could use PeerSync to copy over initial bulk data set to the staging server, convert it, then use PeerSync to keep the files up to date. Another big plus was being able to use PeerSync for bandwidth throttling – which was easy to do. We previously used a piece of throttling software that didn’t support 64 bit machines. All of our servers save one are 64 bit; all of our PCs are 32 bit. Being able to communicate and transfer data regardless of whether you’re 64 or 32 bits is key. PeerSync didn’t care if it’s 32 of 64, it didn’t care about the type or volume of data – ‘I’m a service, I don’t care’. Plus, its multithreading was really critical.”
According to Rickard, a good piece of software must offer three things: “It needs to be intuitive – if it is supposed to do x, it better do x. Responsive and reliable support is absolutely essential. And you want it to address both your immediate needs and those you can envision five years out – you need to know that if you buy it today it will still be effective in five years.” Rickard paused. “PeerSync is a good piece of software. Very good, in fact.”
|
