<   Previous Post  Why Virtualize -...
SVC 4.3.0 - Space-Ef...  Next Post:   >

Comments (8)
  • Add a Comment
  • Edit
  • More Actions v
  • Quarantine this Entry

1 localhost commented Trackback

You're right, Barry -- our friend Hu has seemed to distance himself from the real world in multiple dimensions.<div>&nbsp;</div> It'd be great to get into a semi-honest, semi-transparent discussion with -- or anyone else at HDS -- but my feeling is that -- since they can't stand the heat -- they don't venture into the kitchen that often<div>&nbsp;</div> Keep after it -- someone's got to keep him honest. And, despite our respective vendor affiliations, I agree with most everything you stated in this post.

2 localhost commented Trackback

Well, I was coming here to comment that I agree with most everything you said as well, but I see that Chuck beat me to it.<div>&nbsp;</div> Being the storage anarchist, I don't think I'm allowed to agree with both of you :)<div>&nbsp;</div> Seriously, Hu's obviously lost contact not only with reality, but also with the mother ship - witness the disparity between his public comments on Flash Drives ("nobody needs them", "they don't support SMART", etc.) and Hitachi's press release last week where they said that they'd be supporting some sort of flash drives from one or more as-yet undetermined vendors, an an unspecified timeframe.<div>&nbsp;</div> Almost as funny as "tape" (sorry, just kidding).<div>&nbsp;</div> Seriously, good points, all. And don't let up on Hu - I promise I won't.

3 localhost commented Permalink

Thanks guys, glad its not just me that finds his posts out of this world!<div>&nbsp;</div> Han,<div>&nbsp;</div> Thanks for the feedback, I passed this on to the team managing the interop and have some comments back.<div>&nbsp;</div> The RPQ process allows users, customers, account teams to request support for particular devices. There is a finite set of resource available to perform such qualification. The more requests for the same product ovbiously influences the order in which things get done. Meanwhile general functional and system test cycles will pull in several new controllers - assuming the respective vendors are happy to sell us their boxes once they know what we are doing with them! We do have a few new boxes in qualification at the moment.<div>&nbsp;</div> We do have the Dell/EMC Clarrion support, however we have not been asked for other Dell support via the RPQ process. SUN rebadges most of their storage products and you will find that most of the OEM vendors are supported - therefore the SUN boxes are also supported. The DS3400 is supported, its the only FC interface so the other products in the 3000 range cannot be attached.<div>&nbsp;</div> As for older boxes, we will not remove support for something thats already supported, the list only grows, never gets smaller! In todays financially challenged global environment we are still finding we continue to get a strong and steady stream of requests to support older products, and SVC provides these customers with a way to keep these things working effectively and economically for a longer time. We suspect this trend will continue for some time.

4 localhost commented Trackback

So do you support thin provisioning yet? I read your blog when you post, but I don't remember hearing anything about thin provisioning coming out yet...

5 localhost commented Permalink

&gt;So do you support thin provisioning yet? <div>&nbsp;</div> I am not an IBMer, just a customer. <div>&nbsp;</div> But from what I know SVC can now create what it calls 'space efficent' vdisks which are thin provisioned volumes. You can also now do space efficient flash copies(snapshots), so your snapshot only takes up the space to store the differentials. <div>&nbsp;</div> This was all added in 4.3.0 I believe.<div>&nbsp;</div>

6 localhost commented Permalink

IBM has not yet made any official press release statements about SVC 4.3.0, so in answer to OSG's question, no not yet.<div>&nbsp;</div> PS. Han, sorry to delete your comment, feel free to email me : barry.whyte (at) uk.ibm.com and I will explain.

7 localhost commented Permalink

So your comment "SVC has perfectly linear scaling as you add more nodes, so the workload scales." What happens after the eighth node? You then need to create a seperate cluster correct? Is that not equivelent to adding another USP-V.<div>&nbsp;</div> Can you seemlessly move data from 1 node pair to another node pair?

8 localhost commented Permalink

True if you have performance needs beyond what an 8-node cluster can sustain then you do need to implement another cluster. Point taken, however what I was saying is that you can scale a single cluster from mid-range to beyond that of todays enterprise controllers.<div>&nbsp;</div> Within a cluster moving a vdisk between IO groups means that the WWPN's that are presenting the vdisk change, so you have to reconfigure your multi-pathing software and there is a short period of time where you have to halt your applications. But you can move vdisks using the chvdisk command.<div>&nbsp;</div> Between clusters you can use the Metro or Global Mirror functions to migrate vdisks.