I’m Glad I’m Not The Only One

It would appear that Bob Plankers of LoneSysAdmin and I are thinking along the same lines.  Shortly after I posted a brief notice of the Storage VMotion GUI, Bob posted a rant on the overall quality of VirtualCenter 2.5.  In Bob’s words, I don’t even have to post, because he said exactly what I was going to say.

I didn’t hesitate to proclaim my disappointment in the Remote CLI, but I had hesitated to post my disappointments regarding other aspects of version 2.5 of VirtualCenter.  It’s slower than earlier versions on the same hardware, and overall it feels…unfinished.  Almost like VMware rushed to get this out the door too quickly, perhaps as if they were feeling the pressure from Citrix/XenSource, Microsoft, and Virtual Iron and felt like they absolutely must get this software out before it’s completely done.  After reading Bob’s post, I’m glad to see I’m not the only one feeling this way.

It’s all through the product.  VMware introduces support for jumbo frames, but only for virtual machines—not for the VMkernel to use with iSCSI or NFS datastores, where it could really help.  VMware gives us Storage VMotion, but hobbles it with the Remote CLI.

C’mon, VMware, there’s no need to rush things.  Organizations use your products because they are rock-solid and they provide the essential functionality required.  Focus on the solidity.  Capitalize on existing strengths.  Shore up weakness, but don’t introduce new weaknesses during the process.  Only then will you be able to fend off Microsoft’s “good enough” hypervisor bundled with every copy of Windows Server 2008 Enterprise.

So it’s clear that Bob and I aren’t too terribly happy with the overall fit and finish of VMware Infrastructure 3 version 3.5.  It’s a good product, yes, but it could have been a great product.  Now it looks like we’ll need to wait until version 3.5.1 until we can expect that.

What about you?  What do you think?

Tags: , ,

  1. Joe Cruz’s avatar

    Hey, Scott.

    You know, before you and Bob mentioned it, I couldn’t really put my finger on that nagging feeling I got after upgrading to VC 2.5/ESX 3.5. The “fit and finish”, as you put it, is definitely slightly marred in this latest version.

    Here’s hoping that VMware shakes themselves out of whatever funk is affecting their development!

    In the meantime, I’m going to be doing some due diligence discovery on the Hyper-V products as Windows 2008 hits the shelves.

  2. vmzare’s avatar

    Thanks for the info and fully agree with your suggestion to VMWare in general.

  3. hot carl’s avatar

    similar to joe cruz (above), i had a “nagging feeling” about vc 2.5/esx 3.5. fortunately i got that nagging feeling *before* doing any upgrades. unfortunately our upgrade plan was already in motion and a co-worker upgraded to vc 2.5. i managed to convince everyone to hold, so we haven’t yet went to esx 3.5. as a 10+ year linux user, i *really* like the service console and hate to see it go.

    luckily for me (us) i stumbled across some of you and bob’s posts and have been following both closely. we’re now not planning on upgrading to esx 3.5 anytime soon.

    thanks for the posts and keeping us all updated!

  4. Dmitry Sorokin’s avatar

    It’s possible to configure jumbo frames for VMkernel interfaces. There’s nothing in the GIU, but it all can be done with esxcfg-* commands. We did that in our LAB and did some IO tests, but didn’t see any performance improvements. Probably because of the nature of the test (all random IO). With sequential IO it might help, but that type of IO you rarely see in real world.

    Dmitry

  5. peter.e’s avatar

    hi scott

    you are absoluteley right in a few points. jumbo-frames not “supported” in the vmkernel or better “experimental-status” which is a word i dont want to read in “rocksolid” RTM`d enterprise software- release notes made me also think about the leading virtualization player.

    somehow i have the same feeling than you, ESX3.5 RTM`ed exactly at christmas, to late if you are looking back. i double that: its unfinished in some aspects so absoluteley no ESX4 right now.

    But i cant get the point regarding the remote-cli: IMHO it takes much longer (in the case of app deployment) providing a cli for windows+linux + a virtual-appliance etc. than programming a simple gui, just a window to VC. i have learned from my trainer that vmware doesnt think you will use this feature in the daily operating, so there is simple no need for a gui out there in vmwares opinion which is my opinion too of course. but if the customers say: hey we need a gui for that be sure, vmware implements it. i remember back to ESX2.x, everybody wanted to see iSCSI implemented into the vmkernel. now its standard having iSCSI and NFS in VI3. What i want so say: vmware will react on the customers needs.

    i am not related to vmware :)
    regards, p.

  6. slowe’s avatar

    Dmitry,

    You’re absolutely correct–you *CAN* configure jumbo frames on the VMkernel interfaces. In my own tests in the lab, enabling jumbo frames on the vSwitch, enabling jumbo frames in the physical switch, and enabling jumbo frames on the IP storage device resulted in a complete loss of connectivity. That’s right–it just plain didn’t work. At least yours worked. :)

    Peter.E,

    You do make a good point about VMware responding to customers calling for software iSCSI and NFS. But now that they are an “enterprise” software vendor, will they still be as responsive to customer requests? Will they still be as concerned about product quality?

    Hot Carl,

    Keep in mind that ESX 3.5 doesn’t do away with the Service Console–that’s ESX 3i, the “embedded” version. “Regular” ESX 3.5 still has the Service Console.

    It’s really more VirtualCenter quality than ESX quality, though, if you think about it.

  7. Pkillur’s avatar

    So, I have quite a few friends who work at vmware, who was there when it RTM’d there were quite a few things that got kluged. One thing I haven’t seen a lot about was the fact that they dropped support for certain Lefthand iSCSI arrays. It seems like there were a gajillion “my LH array went kaput” cases.

    I DO think that they’ll fix the issues though, they were racing around the clock (literally) to fix a few of the bugs.

    If it makes anyone feel any better, most of the support guys thought the RCLI blows too… :)

Reply

Your email address will not be published. Required fields are marked *

You may use these HTML tags and attributes: <a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>