by dave on August 6, 2008

Which memory locations can be cached by which ...Image via Wikipedia

Is it just me or have we been waiting for this for a while? 😉  Today is officially the EMC Clariion CX4 public GA (general availability) date.  Good news: they’re shipping TODAY!  No paper launches, folks…this is immediate availability.  The other good news: you get to do more with your storage; faster, cheaper, stronger, more flexible, etc.  Let me rip through some highlights for you:

a.) Cache and SP Processor increases.  Across the board, processor “speeds” and cache sizes have been increased.  Now, this may appear somewhat odd in that the CX4-120, for example, only has two dual core 1.2ghz processors, but, when you consider that the onboard L2 cache is greater in size (and Woodcrest processors were HANDILY more powerful than the older Nocona Xeons), it actually has more innate processing power than the previous generation processors.  Cache sizes, when coupled with the 64 bit FLARE OS for the array, allow for better allocation and utilization within the array. 

b.) Storage capacity.  Oh yes, this is great news to all.  So, essentially the CX4 model nomenclature accurately depicts physical disk counts per array.  CX4-120 = 120 drives, CX4-240 = 240 drives, CX4-480 = 480 drives, CX4-960 = 960 drives.  Not to shabby across the line, right?  Additionally, the actual physical drives themselves are getting a requisite “bump” in capacity (on the Fibre/SAS side of the house, that is).  450Gb Fibre and SAS drives should be available around October in 15K speeds (capacity AND IOP presence; can’t beat that!)

c.) Updated FLARE OS.  FLARE 28 is being introduced with the CX4 line.  It is a true 64 bit OS for the array and, as stated previously allows for better cache allocation and utilization.  Additionally, it paves the way for other “fun” features such as Virtual Provisioning (aka Thin Provisioning), policy-based spin-down for low power drives, and SSDs.  (Sorry Hitachi, we’re actually 2 steps ahead of you now. 😉 )  Note: SSDs aren’t contingent on a 64 bit operating environment, it’s just lumped in there for space concerns.

d.) Low power drives.  Now, this isn’t something that a lot of people have historically paid attention to.  Essentially, by tweaking the motor on a standard 7,200 rpm SATA drive to run at a slightly slower 5,400 rpm spindle speed, you can reduce the power footprint dramatically.  Oh, couple that with our policy-based spin-down “software” in FLARE 28, and you’ve got a “Tier 3” (read: archive, B2D) storage platform within your Clariion.  This has already been introduced on the EMC Disk Library line and now will be going to the mainstream arrays starting with the CX4.  Note:  anyone else seeing the convergence trend here? anyone?

e.) SSDs.  Yes, they merit their own mention.  Remembering that the SSDs (Solid State Drives based on SLC NAND chips) were first introduced @ EMC as part of the Symmetrix line, it should come as no suprise that, *gasp* they’re now going to be available on the CX4 as well.  Well, not through the entire line (yet) but, starting with the CX4-960 and CX4-480s, you’ll be able to hit a threshold of 98% greater efficiency per IOP than any competing disk technology.  Same rules apply as on the Symmetrix: dedicated backend loop for the SSDs with no other disk technology in the same backend loop.

f.) Ultraflex.  Ultra-WHAT?  So, when the PCIe (PCI Express) standard was created, it allowed for (provided for, etc.) the ability to “hot add” components into a given system without (and this is the “hot” part) powering down. The benefit to the storage industry with this introduction is that, unlike Compellent and NetApp, you don’t have to bring your storage array down to “add” additional Front End or Back End ports.  Further, let’s say that FCOE (Fibre Channel over Ethernet) actually takes off and hits a price point that most Commercial businesses can afford (not likely in the short term…);  want to add those connectivity ports? Simply purchase and plug in.  Same goes for 10GbE, Infiniband (on a Clariion? 😉 ), and 8Gb/s fibre channel.  The additional benefit to this type of architecture is that port failures no longer require ripping the entire Storage Processor from the box to remove a daughterboard.  You can simply rip the offending Ultraflex board out and replace it with a new one (this is my thinking, not engineering, btw, but it “makes sense”).  Again, this is a leg up (or several rungs on the storage ladder) above our competition.

g.) Every unit is a combo unit. With the introduction of the CX4, every single model in the CX4 line will feature both iSCSI and Fibre channel connectivity. With the pervasiveness of iSCSI in the VMWare world, it only made sense to allow for iSCSI to be present on every line item.  Same as with the CX3 line, these iSCSI ports are hardware-TOE based, so, performance will be exemplary.

Anyhow, I’m still on vacation, so, let me beg off any further details for now. 😉  Will write more later.


Zemanta Pixie