Diagnostic and Recovery Tools at the Ready
Reading Paul Thurrott’s “What You Need to Know About Microsoft Desktop Optimization Pack 2009 R2” (April 2010, InstantDoc ID 103602)—as well as Rhonda Layfield’s article “XP to Windows 7 Migration with Microsoft Deployment Toolkit 2010” (April 2010, InstantDoc ID 103607)—inspired me to download and install both products. I enabled the Pre-Execution Environment (PXE) on the server running MDT 2010, and everything worked like a charm. Then I came up with an idea: Why not see if I can install the Diagnostics and Recovery Toolset (DaRT) so that the recovery tools are always at my disposal via PXE boot?
After installing the Microsoft Diagnostics and Recovery Toolset, I browsed to Start, Programs, Microsoft Diagnostics and Recovery Toolset, ERD Commander Boot Media Wizard. I created a boot disk and downloaded the latest Sweeper definitions. I inserted my new DVD into my PXE server. I opened Windows Deployment Services console, right-clicked Boot Images, selected Add Boot Image, and selected the WIM file located on the DaRT DVD.
Now, when I press F12 and boot from the network, I have all the DaRT tools instantly available! I can even walk a user through the repair steps, if necessary.
Failover Clustering Hassles
I'd like to add a remark to John Savill's article, "4 Failover Clustering Hassles and How to Avoid Them" (April 2010, InstantDoc ID 103534). Describing high availability with virtualization, John writes that in case of a virtual server crash it would take some time to restart a new copy of a virtual machine (VM) on another virtual server and so "with unplanned server downtime, you’ll have a period of unavailability." My point is that this is the case for Microsoft's virtualization solution.
VMware vSphere has a special feature called "Fault Tolerance" which allows two copies of a VM to run on different virtual servers simultaneously and to keep their states in sync (http://www.vmware.com/products/fault-tolerance/). This solution provides zero downtime in case of a virtual server failure. I think this fact is worth mentioning since you sometimes write about VMware products in your magazine.
Fault tolerance is a feature in VMware ESX 4 with their high-end versions. However, it’s important to remember that it currently works only with a single-processor VM and has some pretty significant network requirements, as all those CPU instructions that are in lock step have to be sent over the network before being actioned. Thanks for pointing this out, though.
Migrating to Exchange Server 2010 … Not!
The fact that you've devoted a number of articles (and practically the whole May issue) to migrating Exchange has solidified my decision to move away from the Microsoft Money Churning Machine. The sheer cost and amount of work necessary to upgrade are enough to make my head spin. We'll look toward the heavens—or should I say the clouds?—for our future email needs.
We've decided to move to Google Apps. The move makes sense for an organization our size, and the migration is virtually painless. We have only about 100 mailboxes to move. With so many users already using some sort of web-based email (e.g., Gmail, Yahoo!) for their personal email, the learning curve is short and is more focused on how Google Apps differs from Outlook.
We've formed an eight-member test team utilizing dual delivery whereby email is delivered to Exchange and Google Apps simultaneously. This team ensures that we don't miss anything during our testing period. Reports from the team are positive so far. After a few training sessions, we'll be ready to roll out the product and retire the Exchange server.
Why a SAN?
Michele Crockett’s editorial, "Exchange Upgrade Creates Domino Effect" (May 2010, InstantDoc ID 104627), has some good points to consider when upgrading to Exchange 2010. I enjoyed it until I got to the section about the company upgrade.
Why is the company going to a SAN? One of the best features of Exchange 2010 is that you don't need a SAN and can use much cheaper disks and have three or four cheaper trays of drives for the price of one SAN. If all your data is stored on one SAN, you have a single point of failure. Is the company going to have multiple SANs (a very expensive option)? It seems as if the business's old architecture is how the new system should be set up, and the new setup is how their old email system should have been running.
Crockett offers good points about the necessity to upgrade Outlook. But remember that Outlook Web Access (OWA) is very robust and has about 95 percent of Outlook's functionality! Keep up the good work.
Shrink an NTFS Volume
I'm responding to John Savill’s FAQ, “How can I check the amount that I can shrink a volume from the command line?" (April 27, 2010, InstantDoc ID 125145). I'd like to give my input on the subject of shrinkable space available.
I had the same problem: lots of free space, but the shrinkable size was far less than the amount free. So, I defragged and defragged, but I saw no change in the space available to shrink. Then I came up with a solution: Before performing the shrink operation, move the page file during the operations, then turn off System Restore. Suddenly, I had lots of free shrinkable space. When the shrinking operation was done, I moved the page file back and turned on System Restore—on my more conveniently sized system drive.
For more detail on this topic, check out John Savill’s FAQ, “I'm trying to shrink an NTFS volume, but the shrink value possible is far less than my free space. What's wrong?” (April 26, 2010, InstantDoc ID 125144)
Microsoft vs. Google in the Cloud
After reading Paul Thurrott’s commentary, “Kickin’ It in the Cloud” (May 18, 2010, InstantDoc ID 125256), I'm inclined to believe that Paul doesn't deal with Microsoft from a technical standpoint very often. I would trust something that Google puts into beta much more than I would trust a “completed” Microsoft product. The first commercial edition of Windows Vista should have been the beta. If you compare Vista and Vista SP1/SP2, the original was the equivalent of a poor beta that users pumped millions of dollars into before Microsoft figures out how to make it work effectively. And what about Microsoft’s several failed attempts to provide a 64-bit OS? And how the company tried to force its customers to use 64-bit via OEM manufacturers and gave no warning of software/ hardware compatibility issues? Paul also doesn't mention how Microsoft has attempted to recreate nearly all of Google’s products.
In the May issue's Community Forum, Brett A. Bennett wrote, “Raid 5 is slightly slower with writes only because the disks must skip over the parity blocks”. The text should have read, "Raid 5 is slightly slower with reads …" We apologize for the error.