* HyperIP block-level compression technology from NetEx
Most of us have suffered at the hands of a vendor at one time or another. In fact, it’s probably fair to say that most of us have suffered at the hands of several vendors several times.
The result is that experienced IT managers have learned to be a bit gun-shy when it comes to relying on a single company to provide a particular product or service. On the one hand we like the idea of being able to consolidate our buying and hopefully squeezing some additional discount out of each transaction; on the other hand, we also like to have a reliable second source available, just in case.
How then do we explain companies that have gone with a single vendor despite the assumed dangers of such a decision? Moreover, what about those intrepid few who have not only worked with a single provider of some important technology,but also go with a proprietary implementation or product set?
IP-based storage networks have been around for a while now. Originally, they came on the scene as a midrange solution for shops that couldn’t (or wouldn’t) invest in Fibre Channel. Now, however, storage vendors are readying gigabit Ethernet implementations for enterprise-class storage arrays, and EMC already has one in the marketplace. And of course, many midrange devices offer IP interfaces as alternative to Fibre.
Typically, the data transport between array and application is via TCP/IP, which – if I recall properly – has been commercially available since the mid-1980s. A problem with TCP/IP however is that as distances increase and as payloads get larger, performance degrades. Remotely replicating data across large distances via TCP/IP can suffer due to packet loss, as of course would the I/O from even the best filer when it is called upon to do long-haul transfers.
All this required bandwidth of course costs big bucks, and if packet loss reduces throughput to less than 30% of its potential, then bandwidth becomes a much less-useful measuring tool than we had assumed and the actual cost of long-distance I/O may turn out to be three times what we thought it was.
If this is an issue for you, a solution to consider is HyperIP from NetEx (http://www.netex.com). Its block-level compression technology, which puts HyperIP boxes between the IP switches at each end of the transfer (two at each end, if you require a high availability solution), claims to triple throughput while remaining transparent to both applications and networking infrastructure. Connect the device to the Ethernet switch and it is ready to go.
The device is qualified with a number of industry-standard products, including flagship vendors such as EMC, IBM Tivoli and Veritas. But still, it is proprietary. Who would trust such a thing?
It turns out that NetEx and its products, in one form or another, have been around for quite some time. Its customer list includes blue chip accounts in government, transportation, telecoms and finance on four continents. I was surprised to find out that I had been using its technology 20 years ago on Prime computers.
If you use TCP/IP to transport data for distances of more than 50 circuit miles and have throughput that is less than 90% of your bandwidth, it may make good sense to check out NetEx.
