Is there a justifiable need for HP Flex 10?
HP recently announced their new flex10 virtual connect modules for the cClass line of blade enclosures. With the flex10 and the latest G6 module blade servers you could leverage some great technology. Two of the main reasons for me looking at flex10 was the functionality of taken the default two onboard network ports and making them into eight, virtualizing the enclosure, and 10gigE. Other features of the flex10 can be found here.
The benefits for flex10, G6 cClass blades and Vmware Infrastructure 3 (VI3) were odvious on the network side so I had to test them in our lab. The lab setup currently consists of four Cisco 3020’s and two fiber passthru modules. More detailed information can be found in one of my previous posts here. And thanks to HP I was able to get some hardware to test with which consisted of 2x flex10 modules, 2x virtual connect fiber modules, and 2x bl495c blades.
With the new hardware setting in the lab we shutdown the blades and replace the cisco modules and fiber passthru’s with the virtual connect modules. This took us from six i/o modules to four. The virtual connect modules discovered all the blade hardware, we updated the firmware, then configured the modules to fully virtualize the enclosures mac addresses and world wide numbers (wwn). I will not go into detail but it was not rocket science but it did take us some time. There were a few issues but for the most part they were all resolved after updating the bios on the older G4 blade servers.
So with more good than bad results with the new flex10 and virtual connect I got some pricing on the configuration. Our current pricing comes in around $32K for the cClass enclosure with all the fans, power supplies, etc (no blades included). The pricing for the same configuration but with 2x flex10 and 2x virtual connect fiber modules replacing 4x cisco 3020 switches and the 2x fiber passthru’s came in over twice as much at around $64K. I expected it to cost more but I didn’t think it would be by that much but to HP’s defense the two scenarios are far from apples to apples. Now when comparing the flex10 modules to the cisco 3120x switch modules the pricing makes sense.
So, to answer the question “Is there a justifiable need for HP Flex 10?” I’d have to say “Yes” but not in my case right now. A few reasons would be that:
- We are not ready for 10gigE and don’t have a true need for it yet (not many of our servers use all 1gig of network bandwidth or need 8 nic ports)
- The total solution costs are much higher than the gain we’d receive at this point in time
- The solution does not provide much benefit for our older blade systems (can’t get 10gigE or eight ports from the default onboard nics)
- Have to invest in all new blade servers to take full advantage of flex10
- Our network infrastructure is not completely built out to support 10gigE on the edge
Justifying IT spending is hard enough these days for many companies so we have to squeeze all the life out of what we have before spending money on the latest and greatest unless we can quickly see a return. I do like the products though and plan on looking at this again six months out. Hopefully pricing would have gone down by then and our datacenter needs have changed as well.