January 2011

You are currently browsing the monthly archive for January 2011.

Announcing VMware vSphere Design

Monday, January 31, 2011 in Collaboration, Virtualization by slowe | 27 comments

This is a joint post of three prominent writers in the virtualization community: Forbes Guthrie, Scott Lowe, and Maish Saidel-Keesing.

For the past 6 months we have been working on a joint project. It has actually been kept pretty quiet, and it is now time to make this public.

Previous VMware vSphere books have focused on how to master the technology, deep-diving into certain elements and giving tips & tricks that help you manage your virtual infrastructure.

But we felt there was something missing in all these books. What was missing was how to design an infrastructure, accounting for all the elements that you need to consider. For example:

What kind of servers should I use?
Which storage: NFS, iSCSI or FC?
How do I scale a vCenter Server appropriately?

…and many more similar questions. This is how VMware vSphere Design was born…

The three of us collaborated on the book, to not only explain how to configure each element of your infrastructure, but to make you think about all the options available, and how each choice can impact the overall design. It should help you find the right solution for your environment—because no “one size fits all.”

To our knowledge, it is the only book focused on designing VMware vSphere implementations. It is written for engineers and architects who plan, install, maintain and optimize vSphere solutions.

The book details the overall design process, server hardware selection, network layout, security considerations, storage infrastructure, virtual machine design, and more. We debate the merits of scaling up servers versus scaling out, ESX versus ESXi hypervisors, vSwitches versus dvSwitches, and FC, FCoE, iSCSI or NFS storage. We show you which tools can be used to monitor, to plan, to manage, to deploy and to secure your vSphere landscape. We run through the design decisions that a typical company might face, and question the choices you come to. The book is packed with real-world proven strategies. VMware vSphere Design examines how the virtualization architecture for your company should ideally look, be it a newly deployed environment or an optimization of the existing infrastructure.

VMware vSphere Design is available now for pre-order on Amazon and will be in the stores around the middle of March 2011.

We would like to thank Jason Boche for acting as the technical editor for the book. (Note from Scott: I can’t speak for Forbes or Maish, but I know that Jason’s contributions as technical editor definitely improved the quality of this book. His role should not be underestimated!)

We hope you enjoy reading this book as much as we enjoyed writing it!

Tags: Virtualization, VMware, vSphere, Writing

Presentation on VMware vSphere 4.1 Storage Features

Saturday, January 29, 2011 in Storage, Virtualization by slowe | 1 comment

I had the privilege this past week of speaking at the Triad VMUG meeting in Greensboro, NC, along with fellow blogger, podcaster (is that a real word?), and speaker Rich Brambley. In the event you were unable to attend (there was quite a crowd attending), here’s the presentation I gave at the meeting.

Feel free to share any additional thoughts, questions, or comments below. Thank you!

Tags: Storage, Virtualization, VMware, vSphere

Technology Short Take #10

Thursday, January 27, 2011 in Networking, Storage, Virtualization by slowe | 4 comments

Welcome to Technology Short Take #10, my latest collection of data center-oriented links, articles, thoughts, and tidbits from around the Internet. I hope you find something useful or informative!

Networking

Link aggregation with VMware vSwitches is something I’ve touched upon a great many posts here on my site, but one thing that I don’t know I’ve ever specifically called out is that VMware vSwitches don’t support LACP. But that’s OK—Ivan Pepelnjak takes care of that for me with his recent post on LACP and the VMware vSwitch. He’s absolutely right: there’s no LACP support in VMware vSphere 4.x or any previous version.
Stephen Foskett does a great job of providing a plain English guide to CNA compatibility. Thanks, Stephen!
And while we are on the topic of Mr. Foskett, he also authored this piece on NFS in converged network environments. The article seemed a bit short for some reason. It kind of felt like the subject could have used a deeper, more thorough treatment. It’s still worth a read, though.
Need to trace a MAC address in your data center? CiscoZine provides all the necessary details in their post on how to trace a MAC address.
Jeremy Stretch of PacketLife.net provides a good overview of using WANem. If you need to do some WAN emulation/testing, this is worth reading.
Jeremy also does a walkthrough of configuring OSPF between Cisco and Force10 networking equipment.
I don’t entirely understand all the networking wisdom found here, but this post by Brad Hedlund on Nexus 7000 routing and vPC peer links is something I’m going to bookmark for when my networking prowess is sufficient for me to fully grasp the concepts. That might take a while…
On the other hand, this post by Brad on FCoE, VN-Tag, FEX, and vPC is something I can (and did) assimilate much more readily.
Erik Smith documents the steps for enabling FCoE QoS on the Nexus 5548, something that Brad Hedlund alerted me to via Twitter. It turns out, as Erik describes in his post about FCoE login failure with Nexus 5548, that without the FCoE QoS enabled fabric logins will fail. If you’re thinking of deploying Nexus 5548 switches, definitely keep this in mind.

Servers

In the event you haven’t already read up on it, the UCS 1.4(1) release for Cisco UCS was a pretty major release. See the write-up here by M. Sean McGee. By the way, Sean is an outstanding resource for UCS information; if you aren’t subscribed to his blog, you should be.
Dave Alexander also has a good discussion about some of the reasoning behind why certain things are or are not in Cisco UCS.

Storage

Nigel Poulton tackles a comparison between the HDS VSP and the EMC VMAX. I think he does a pretty good job of comparing and contrasting the two products, and I’m looking forward to his software-focused review of these two products in the future.
Brandon Riley provides his view of the recently-announced EMC VNX. The discussion in the comments about the choice of form factor (EFD) for flash-based cache is worth reading, too.
Andre Leibovici discusses the need for proper storage architecture in this treatment of IOPs, read/write ratios, and storage tiering with VDI. While his discussion is VDI-focused, the things he discussed are important to consider with any storage project, not just VDI. I would contend that too many organizations don’t do this sort of important homework when virtualizing applications (especially “heavier” workloads with more significant resource requirements), which is why the applications don’t perform as well after being virtualized. But that’s another topic for another day…
Environments running VMware Site Recovery Manager with the EMC CLARiiON SRA should have a look at this article.
Jason Boche recently published his results from a series of tests on jumbo frames with NFS and iSCSI in a VMware vSphere environment. There’s lots of great information in this post—I highly recommend reading it.

Virtualization

What, you didn’t think I’d overlook virtualization, did you?

I like the idea of CLASH. It just needs more development.
Running VMware Fusion? I am. Ever needed guest VMs on Fusion to have a static (fixed) IP address? I have. Here’s how it’s done (hint: Fusion uses a hidden DHCP server with its own configuration file).
This VMware KB article tells you how to change the default password hashing mechanism from MD5 to SHA512.
If you haven’t yet upgraded to vSphere 4.1, this VMware KB article is a nice supplement to the vSphere Upgrade Guide.
In Technology Short Take #9, I referenced Part 1 and Part 2 of Frank Denneman’s series of articles on the impact of oversized virtual machines. Since then Frank has published part 3, in which he looks at oversized virtual machines from the perspectives of CPU scheduling, memory management, and bootstorms.
I also liked Frank’s article on setting cluster resource percentages correctly. That’s a “little thing” that people often overlook, I think.
Oh, and one other thing: Frank, I think you can stop beating the CPU affinity dead horse now, OK?
Ben Armstrong posts a couple of useful Hyper-V posts (off-topic question: should he change his handle from “Virtual PC Guy” to “Hyper-V Guy”?). First is a discussion of how Hyper-V handles VM shutdowns, followed by a post on how to use PowerShell to determine which physical computer you’re running on.
Duncan has a great post on the interaction between SIOC (Storage IO Control) and “external” workloads. This is something that both Scott Drummonds and I discussed earlier (see my post and see Scott D’s post), but it’s good to see it get some additional attention. Duncan and I traded some comments and some tweets about the issue, particularly with regard to the interaction of SIOC and low-RPO replication solutions. I’ve got some thoughts on that interaction that I need to work through, but don’t be surprised to see a blog post on that in the near future.
Hany Michael has a tutorial on getting vCloud Director working with vShield Edge and vShield App.

Before I wrap up, I’ll just leave with you a few other links from my collection:

IOBlazer
Backing up, and restoring, VMware vCloud Director provisioned virtual machines
RSA SecurBook on Cloud Security and Compliance
Hyper-V Live Migration using SRDF/CE – Geographically Dispersed Clustering
The VCE Model: Yes, it is different
How to make a PowerShell server side VMware vCenter plugin
VMware vSphere 4 Performance with Extreme I/O Workloads
VMware KB: ESX Hosts Might Experience Read Performance Issues with Certain Storage Arrays
vSphere “Gold” Image Creation on UCS, MDS, and NetApp with PowerShell
Upgrading to ESX 4.1 with the Nexus 1000V
My System Engineer’s toolkit for Mac

That’s going to do it for this time around. As always, courteous comments are welcome and encouraged!

Tags: Cisco, EMC, HyperV, Microsoft, NetApp, Networking, Storage, Virtualization, VMware, vSphere

Has Virtualization Really Stalled?

Wednesday, January 26, 2011 in Virtualization by slowe | 44 comments

Has virtualization truly stalled? It’s an interesting question. What got me thinking about this topic was this InfoWorld article. Quoting from the article:

One big reason: Server virtualization has stalled. Research shops and vendors have published numbers indicating that although most enterprises have stuck their toe into the waters of virtualization, they’ve hardly plunged in.

If virtualization has truly stalled, my question is this: why? What are the reasons, misperceptions, or other drivers that are causing organizations not to adopt virtualization more broadly? Do customers think that virtualization—and here I’m thinking specifically of VMware vSphere, as the market leader—can’t perform well enough for mission critical workloads? That it isn’t reliable enough?

I have my own opinions, but I’m really interested in hearing from the community. Please share your constructive thoughts in the comments below. As always, please be sure to disclose affiliation where necessary. Thanks!

Tags: Virtualization, VMware, vSphere

Clearing Up the Bad Reputation of VMware’s Paravirtualized Drivers

Thursday, January 20, 2011 in Virtualization by slowe | 13 comments

In 2009, not too long after the release of VMware vSphere 4.0, I blogged about the use of PVSCSI and VMXNET3; specifically, I mentioned reasons not to use PVSCSI and VMXNET3. A lot has changed since then, so—prompted by a reader who shall remain nameless but knows who he/she is—I thought it might be prudent or useful to post a brief update.

While vSphere 4.0 did not support either PVSCSI or VMXNET3 for use with VMware Fault Tolerance (FT), those restrictions were lifted with the release of VMware vSphere 4.1. This was mentioned a couple of times in the comments to the original article, but I did want to clarify it so that readers knew for sure. For more information, see pages 35 and 36 of the VMware vSphere 4.1 Availability Guide. PVSCSI is not explicitly called out there, but it also isn’t mentioned; VMXNET3 is specifically called out as supported with VMware FT.

In addition, the recommendation against using PVSCSI with low I/O workloads was also removed with vSphere 4.1. See the brief note in the Solution section of this VMware KB article.

Feel free to post any corrections or clarifications in the comments. In particular, if you have links to articles or documents with explicit mention of support for these paravirtualized drivers, feel free to share them for the benefit of all readers. Thanks!

Tags: Networking, Storage, Virtualization, VMware, vSphere

Additional RecoverPoint-VAAI Update

Thursday, January 20, 2011 in Interoperability, Storage, Virtualization by slowe | 2 comments

In late October and early November 2010 I published a couple of articles on interoperability between EMC RecoverPoint and VAAI (vStorage APIs for Array Integration, part of VMware vSphere 4.1). If you’d like to go back and read those articles for completeness, here are the links:

RecoverPoint and VAAI Interoperability
RecoverPoint and VAAI Update

In the comments to the second article, a reader indicated that he’d seen a problem when using RecoverPoint (which I’ll abbreviate as RP from here on) and VAAI. In this particular situation, his consistency groups (CGs) were failing to initialize. The only way he could get his CGs to properly initialize and replicate was to disable VAAI. I provided his information to RP product management, who after some additional testing found that there was indeed a potential issue when using hardware-assisted locking (also referred to as CAS, after the SCSI command Compare and Swap) in conjunction with the FLARE 30 array splitter and VAAI.

The fix for this potential issue is found in a type 2 patch for FLARE 30; this patch brings the FLARE 30 version to 4.30.000.5.509.

<aside>For those of you that don’t know, type 2 patches are patch revisions that run through full qualification cycles and have formal releases. These are the sorts of patches that you should install when you can in order to stay current.</aside>

If you are running a revision of FLARE 30 prior to 509, you could see issues when using the RP array splitter and VAAI, and you will need to disable VAAI in order to resolve the issues. With the latest revision of FLARE 30 (4.30.000.5.509), this RP-VAAI interoperability issue is resolved.

If you have any questions, please feel free to ask them in the comments below. Thanks!

Tags: EMC, Interoperability, Storage, Virtualization, VMware, vSphere

Changing the IP Address on a VPLEX Management Server

Tuesday, January 18, 2011 in Storage by slowe | No comments

EMC VPLEX leverages a Linux-based management server as an integral part of the overall architecture. The management server, as the name implies, provides the management interfaces—both HTTPS (web-based) and CLI—for managing the VPLEX cluster(s). In a VPLEX Local configuration, changing the IP address is a single-step process; in a VPLEX Metro configuration, there is an additional step required. In this post I’ll walk you through both situations.

All in all, it’s a pretty simple process, but the one thing that might trip you up is the different CLI environments involved. Some commands are run from the management server’s native Linux-based CLI, while other commands are run from the Vplexcli. Remember that you’ll access the management server’s Linux-based CLI by simply opening an SSH session to the management server. You’ll access the Vplexcli by running either the vplexcli or telnet localhost 49500 commands once you’ve logged into the management server.

VPLEX Local

When you are running in a VPLEX Local configuration, changing the IP address of the management server is a single-step process. From the Vplexcli, a single command is all you need:

management-server set-ip -i <IP address:Netmask> -g <Gateway IP address> -p eth3

This command changes the IP address on the management server and you’re done.

VPLEX Metro

In addition to providing the management interfaces for the VPLEX cluster(s), in a VPLEX Metro configuration each management server is also responsible for creating and maintaining an IPSec-based VPN between itself and its peer management server in a VPLEX Metro configuration. The management servers use this VPN as a private connection between them in order to communicate with the directors in the remote cluster. Therefore, changing the IP address of a management server in a VPLEX Metro configuration could impact more than just the address used to access the VPLEX cluster(s).

So, in a VPLEX Metro configuration there’s a second step you must also follow in order to update the configuration for the VPN between the two management servers.

First, you’ll change the management server’s IP address using the management-server set-ip command I described above.

After changing the IP address using the management-server set-ip command, you then need to edit the /etc/ipsec.conf file on the other management server to reflect the new IP address you just assigned. For example, if you changed the IP address on management server 1 to 172.16.100.100, then you would edit the /etc/ipsec.conf file on management server 2 to show that management server 1 is now reachable at a new address. This ensures that the VPN configuration on each management server properly reflects the correct IP address of its peer.

To do this, open an SSH session to the peer management server—whose VPN configuration needs to be updated—and use vi to edit the /etc/ipsec.conf file. The specific line in the file that needs to be changed is the “right=” line, where the IP address indicates the remote server (“right” means “remote” in this case). Save your changes and return to the management server’s Linux-based CLI.

You’ll then need to re-enter the Vplexcli (again, using the vplexcli or telnet localhost 49500 commands). Once in the Vplexcli, restart the VPN using the vpn restart command. Restart the VPN on both management servers using this process.

To verify that the VPN is working properly, from the Vplexcli you can use the vpn status command like this:

vpn status -l <Local cluster ID> -n <Number of engines in local cluster> -c <Remote cluster ID> -e <Number of engines in remote cluster> -r <IP address of remote management server>

The output of that command should show that each director in each engine, both local and remote, is reachable.

If you need to change the IP addresses of both management servers in a VPLEX Metro configuration, then change one at a time following the process I described above.

Tags: EMC, Storage, VPLEX

Looking Back, Looking Forward

Friday, January 7, 2011 in General by slowe | 16 comments

Everyone else is doing it, so I figured I might as well also: publish something about how the site fared during 2010. I’m going to do that, yes, but I’m also going to talk a little bit about my commitments to the site (and to the readers) for 2011.

Looking Back: 2010

The site ended the year with just shy of 1.2 million views during 2010. That’s a pretty impressive number (at least to me), but it’s barely unchanged since last year. I guess I’m going to have to find new ways of driving visitors to my site!

Here are the top 10 articles on the site (these articles could have been published anytime, not just in 2010):

The top 10 articles published in 2010 is interesting as well; I found that all of the most popular articles on the site were published in previous years. I don’t know if this means my content is getting worse (so older content is better than newer content) or if it just means the older content shows up better in search results. Anyway, here are the top 10 articles published in 2010:

The thing I found interesting about this list is that some of the posts I expected to be on there—like some of the FCoE-related posts—are nowhere to be found. Interesting…

Looking Forward: 2011

For 2011, I have a few commitments to the site and to the readers:

One thing that I haven’t done a good job with over the last year or so is responding to readers’ comments. So, this year, I’m committing to do a better job of responding to readers’ comments here on the site. If you post a comment, I’m going to do my absolute best to respond to your comment, even if that means simply saying “Thank you”.
I am committing to continue to provide full RSS feeds and not just summaries. I’m also committing to not include advertisements of any sort within the RSS feeds. That being said, I might end up switching to excerpts or summaries on the home page in order to draw more readers deeper into the site.
A lot of readers have asked for the return of search functionality. So, this year, I’m committing to bring back search functionality to the site.

I do appreciate every single person who visits the site, subscribes to the feeds, or posts a comment. To each and every reader: thank you! I will strive to provide solid, useful, pertinent technical information that will make it worthwhile to continue to be a reader!

Tags: Blogging

New vSphere Resource by Train Signal

Wednesday, January 5, 2011 in Virtualization by slowe | 2 comments

David Davis of Train Signal recently dropped me an e-mail to alert me to the release of a new series of videos titled VMware vSphere Troubleshooting Training. Having watched a fair number of the Train Signal videos—I used them while prepping for my CCNA exam late last year and have watched many of the vSphere videos—I’m confident in saying that this will be a valuable resource for anyone who is interested in learning more about vSphere and how to troubleshoot vSphere. This is especially true for users who are new to vSphere. (I know it’s hard to believe that some users are new to vSphere, especially for the long-time users, but it is very true.)

However, even for users who are a bit more experienced with vSphere, David structured this troubleshooting course around the VMware Certified Advanced Professional Data Center Administration (VCAP-DCA) blueprint and this course will be part of an upcoming VCAP-DCA training package. So, those of you who are out there prepping for VCAP-DCA—keeping in mind that VCAP-DCA is a necessary first step on the path toward VCDX—might be interested in this training course as well.

A snippet of one of the troubleshooting videos is available online in the event you’d like to take a look.

Disclaimer: Train Signal is a paid sponsor of this site and David Davis is a personal acquaintance. However, I received no consideration of any kind, financial or otherwise, for this post.

Tags: Virtualization, VMware, vSphere