I haven't kept this as up to date as I'd like but the stuff that is here is good to know. I'm going to modify it so other people can add sections to it and it can be a living document, but until then this is all you get.
In a SQL Server Magazine podcast, David DeWitt, a Microsoft technical fellow,discusses parallel computing, the topic of his keynote at PASS, and the Jim Gray Systems Lab that he runs at the University of Wisconsin.
Send an HTA greeting card that features a holiday tree with blinking multicolored lights and a "Happy Holidays" greeting. You can easily personalize the greeting.
An often irreverent look at some of the week's other news, including a Vista Capable dismissal request, Zune price reductions, Morrow musings, Novell and Microsoft sitting in a tree ... two years later, Yahoo!, IE 6 on Windows Mobile, and so much more
In this SQL Server Magazine podcast, Quest executives and technical experts discuss the growth of SQL Server in the market, industry trends, SQL Server in the cloud, and the pros and cons of SQL virtualization.
Would you stop doing backups on your Exchange systems if you could? That might seem like a foolish question. For most Exchange administrators, backups are right up there with motherhood and apple pie as core beliefs; I’d venture a guess that very few of us would be comfortable not taking any backups at all. However, Microsoft explicitly supports backup-free Exchange environments in some circumstances. How is this possible? And perhaps more important, is it a good idea?
You might remember an old-school Exchange feature called circular logging. When circular logging is enabled, Exchange keeps a finite number of transaction logs instead of opening a new log-file generation each time the previous log file fills. An advantage of circular logging is that its disk space usage is predictable, but a major disadvantage is that your recovery options are limited because previous transactions are overwritten. For that reason, circular logging is used only in a very small number of circumstances, such as when you need to drastically restrict the growth of transaction logs when moving a large number of mailboxes.
A relatively new Exchange feature is cluster continuous replication (CCR). When you use CCR, log files generated on the active server are replicated by Exchange to the passive server. The replicated log files are played into the passive copy of the database so that the database remains up-to-date. You could certainly argue that after a log file has been replicated and replayed, there’s no need to keep it around—if you have some other protection mechanism, such as a Volume Shadow Copy Service (VSS) point-in-time copy or a vendor-specific clone, to protect the original databases.
Quite a bit of communication occurs between the active and passive nodes to indicate which log files have been copied, which have been replayed, and which can be safely deleted. It’s this mechanism that allows you the option of running without backups: A log will be deleted only if the CCR active and passive targets agree that it’s been properly replayed. In addition, if either the active or passive nodes are a source for a Standby Continuous Replication (SCR) target, the SCR node must have received and replayed the log as well.
Now that we’ve covered the “how” of going backup free, the larger question is whether it’s a good idea to do so. Millions of mailboxes around the world are protected by CCR, including those belonging to all of Microsoft’s employees. It’s a trustworthy technology. Likewise, VSS is a mature solution, and if you combine CCR and VSS-based copies, the odds are quite good that you’ll be able to quickly recover data in the event of a failure. However, the whole concept of not having a purpose-built backup mechanism makes me a bit nervous, sort of like embarking on a long car trip without a spare tire. Sure, there’s AAA and roadside assistance if you need it, but that’s not the same as having your own last-ditch protection.
Microsoft supports the option of skipping Exchange backups, but nowhere does the company recommend it. That’s because the decision to stop performing regular backups isn’t to be taken lightly. If you’re currently running without backups, or considering replacing your backup solution with a combination of CCR and disk-based point-in-time copies, I’d love to hear from you—and if you think this is a crazy concept, that would be good to know too!
Users of Microsoft's Zune Pass subscription currently pay $15 a month to gain access to millions of music tracks via the Zune Marketplace, but until today, that music became unaccessible if they let their subscriptions run out. Now, however, Microsoft
At Microsoft's annual shareholder meeting this week, CEO Steve Ballmer emphatically closed the door on the possibility that his company was interested in purchasing ailing Internet giant Yahoo!. His frank assessment of the situation sent Yahoo!'s stock
Xbox 360 owners who logon to the system's Xbox Live system this morning will receive the most significant functional change yet to the console's user interface, or dashboard. Dubbed the New Xbox Experience, this new front-end features a completely new
Microsoft on Tuesday announced that it would retire its $50-a-year security subscription product, Windows Live OneCare, and replace it with a free solution codenamed "Morro." Unlike OneCare, however, Morro will focus only on core anti-malware features and
Ok, this isn't actually security related but its interesting nevertheless: A nifty third-party tool cracks some Windows 7 files to reveal hidden features previously only available to Microsoft employees.
Something wonderful happened this week: Worldwide, the amount of spam email dropped by roughly two-thirds. How did this happen? A single hosting provider, McColo, was disconnected from the Internet.
Think about that for a second: One hosting company was apparently responsible for up to 66 percent of the worldwide spam generated per day. IronPort, a Cisco subsidiary, estimates that there are about 190 billion spam messages sent per day.
This volume of spam would be impossible, of course, if machines hosted at McColo were actually sending all the spam. Many aspects of modern antispam filtering take into account the origin of the message, including the sender IP address, the sender’s reputation for sending spam, the rate of arrival of messages from the address, and so on. That process makes it more difficult for spammers to operate from a single block of IP addresses.
If you’re familiar with the Folding@home or SETI@home, you already understand the solution spammers have used: Get lots of individual computers to do the work. Spammers have turned to using individual computers—mostly Windows machines—that have been compromised by malware that allows them to be remotely controlled. Groups of these machines, known as botnets, offer a great way to send spam because each individual machine can send messages to various destinations. If you get enough machines in a botnet, you can do all sorts of bad things—apparently including flooding the entire Internet with spam.
McColo’s Internet connection was killed in large measure because of the work of Washington Post blogger Brian Krebs. Krebs worked to gather information about McColo’s operations, then passed that evidence to the upstream Internet providers that connected McColo to the broader Internet. Shortly after Krebs’ initial contact, McColo went dark, resulting in a huge drop in worldwide spam levels.
This is of course good news, but the nature of the modern spam ecosystem means that this drop will probably be temporary. A great deal of spam is sent under the aegis of criminal groups that have access to enough money to quickly reconstitute their operations with another compliant hosting provider. Even a temporary respite is still welcome, though. In the longer term, the IT industry needs to continue to identify effective ways to fight botnet-based attacks, including distributed denial-of-service attacks and spam floods, but that’s a much harder problem to solve.
This month's challenge is inspired by an email message: "I'd like to display user names with a space between the first and last name on our Welcome screens (just because I like the slick look of it)." Do you know how to do this?
The Windows System Assessment Tool encompasses much more than the Windows Experience Index--its command-line capabilities can help with troubleshooting.
Let’s talk about a hot topic from this year’s US presidential election: energy. Both major-party candidates agreed that US energy policy needs some serious improvement, although they differed wildly on the best means of achieving those improvements. The IT industry worldwide has seized on the idea of energy conservation as a selling point, and companies all over are starting to recognize that they can save significant amounts of money by improving the efficiency of their IT operations.
How can you tap into these savings in your Microsoft Exchange Server environment? Simply put, the answer is to turn off as many devices as you possibly can. That’s what virtualization software vendors have been saying for years, but there are certainly ways to save energy other than by turning off the servers themselves. In the United States, the average cost of electricity for commercial use in July 2008 was 11.08 cents per kilowatt-hour (see “Average Retail Price of Electricity to Ultimate Customers by End-Use Sector, by State”). Considering that cost estimate, what does it cost you to run your servers? The answer is a resounding, “It depends.”
The US Department of Energy’s Industrial Technologies Program devised a model, described in “Five Ways to Reduce Data Center Server Power Consumption,” that uses the CPU utilization of the servers as the primary factor for calculating power consumption. This model makes sense because other parts of the system have fairly constant power use. For example, hard disk drives spin all the time, RAM has to be refreshed, and monitors’ power draw is constant as long as they’re on. The formula for calculating power usage is simple:
Pn = (Pmax - Pidle) × n/100 + Pidle
To use this formula, you need to know the maximum amount of power a server can draw (Pmax), which is easy to derive from the rating of its power supply. You also need to know the server’s power draw at idle (Pidle), which you can figure out with an electricity usage meter such as a Kill A Watt. When you have those two factors, you can calculate the power draw for a given level of CPU utilization (n)—and the utilization data is easily available with Windows’ built-in performance monitoring utilities. That makes it easy for you to calculate how much power an Exchange server is using, and thus how much you might save by virtualizing it. This formula doesn’t take into account the use of multiple redundant power supplies, but you can factor those in as needed.
What about disks? Their power usage is fairly constant and relatively small (say, 10 watts in normal operation). This amount might not seem like a big deal, but if you have enough disks—as you will in most Exchange environments—those power costs add up. Microsoft includes disk power consumption as a factor in its latest Exchange Server 2007 Storage Cost Calculator. To perform this calculation, you’ll need to know the power consumption for the disks you’re using and the number of disks you have. Simple multiplication tells you the cost per disk per hour; multiply that by the number of hours the disk will run (there are 8760 hours in a year), and you have the total cost.
You’ll have other associated costs, too, such as the cost of cooling your computer systems. Cooling costs have turned out to be significant in many environments, especially with the wide use of multiprocessor and multicore systems. Bear in mind that a report commissioned by AMD, “Estimating Total Power Consumption by Servers in the U.S. and the World,” says that the growth in server power use worldwide comes mostly from having more servers, not from having higher power use per server. I expect to see lots of effort focused on reducing cooling costs; Intel has already demonstrated some promising results using fresh outside air as a cooling medium, as reported in “Intel's secret weapon: Fresh air,” and other major IT vendors such as HP and IBM are working on related products and technologies.
As usually happens, only days after Microsoft publish its advisory regarding a serious problem with RPC a worm was unleashed to exploit the vulnerability. I hope your systems are patched!
If you’re of a certain age, you might remember the popular 1970 song by Chicago, “Does Anybody Really Know What Time It Is?” That’s this week’s topic in a nutshell: How do servers in a domain know what time it is? And why would they care in the first place? The answer, in a word, is Kerberos, the authentication mechanism used by Windows 2000 and all later Windows OSs. Kerberos depends on time synchronization between the computer that issues Kerberos tickets (the Key Distribution Center—KDC) and the computers that request them. Interestingly, Kerberos doesn’t depend on the time being accurate. As long as the client and server clocks are within five minutes of each other, it doesn’t matter if the actual time or date matches the real world.
However, Windows tries to keep time both accurate and synchronized by providing a built-in Network Time Protocol (NTP) service, W32Time. The operation of this service, on both the client and server, is detailed in the Microsoft article “Basic Operation of the Windows Time Service” if you’re interested. The basic thing to remember is that there’s a single authoritative time source for each forest: The computer holding the PDC emulator role for the forest is treated as authoritative. Member servers and workstations use the domain controller (DC) that authenticated them for time information, and that DC seeks a time source in the current domain or the parent domain.
The authoritative time source for the domain can synchronize to any time source you like. By default, it uses time.windows.com, but you’re welcome to change it if you need a more precise time source. (You can even bring in your own time hardware, such as an HP Z3801A, and get government-level accuracy for a few hundred dollars).
This synchronization mechanism turns out to be really important when it comes to virtualization. For many administrators, DCs seem like a great place to start applying virtualization technology; if you don’t have a huge load of authentication traffic, starting with DCs seems like a reasonable approach. However, it’s important to realize where your DCs are getting their time data from. The clock on virtualized servers isn’t guaranteed to stay accurate over time. Worse still, the default behavior with both Hyper-V and VMware is for the virtualized DC to synchronize time with the physical host computer. If that host isn’t a domain member or it isn’t getting regular time updates, the time on the DC and clients will inevitably skew, and when the skew exceeds the 5-minute window that Kerberos allows, you’ll start having authentication problems.
Microsoft’s recommendation is simple: Don’t have your virtualized DCs synchronize time with their physical host. Instead, you should configure the PDC emulator to synchronize with an outside source. Active Directory MVP Jorge de Almeida Pinto has a detailed blog post explaining how to configure the time synchronization behavior of your PDC Flexible Single-Master Operation (FSMO) role.
If you need to document your Windows servers, try SYDI-Server, the SYDI version for documenting Windows computers. This tool provides detailed documentation—and you can't beat the price. It's free.
Terry Myerson, a corporate vice president at Microsoft, recently made a big change: He left his job managing the Microsoft Exchange Server business and moved to oversee the Windows Mobile team. I had the opportunity to interview Myerson last week about this change and its impact on the Exchange and Windows Mobile businesses.
First, I asked Myerson what he viewed as his biggest success in his seven years with the Exchange team. His reply was simple: “taking the Exchange business from a messaging underdog to a market leadership position. Exchange is now running in 81 of the Fortune 100 companies and is considered one of the most mission critical systems within corporate IT. Over the past seven years, Exchange has earned the reputation of being a rock-solid messaging system and that is something I, and the team of extremely talented people within the Exchange product, are very proud of.” He followed up by saying that the biggest success of Exchange overall is the team that builds the product. “Internally and externally, the Exchange product team is known as a high-performing team, and we take great pride in this,” Myerson said. The Exchange team has a consistent record of engineering excellence, from its early adoption of automated error reporting to its pioneering Best Practices Analyzer tool, so I think Myerson’s enthusiasm here is well justified.
What about the iPhone? I asked Myerson how he responded to the common argument that Windows Mobile is doomed by Apple’s rapid expansion into the mobile device market. He had a strong riposte: Although Apple has sold nearly 10 million iPhones since its launch, Windows Mobile sold more than 18 million licenses during the fiscal year, and Microsoft’s partners introduced more than 30 new Windows Mobile phones to market in 2008.
Myerson also pointed out that the buzz around Google’s Android platform and the iPhone help to raise awareness of smartphones and why they’re better than the simpler phones from vendors such as Nokia and Motorola that now dominate the worldwide device market. I tend to agree that there’s still huge room for growth in the smartphone market; interestingly, Apple CEO Steve Jobs’ remarks during Apple’s third-quarter earnings call claimed that the iPhone outsold RIM’s BlackBerry line in the quarter. The fact that Jobs, who is notoriously cautious about what he says during earnings calls, mentioned this fact certainly makes it sound as if RIM is the real target for Apple’s growth, giving Microsoft a bit more maneuvering room.
Next, I asked Myerson where he thought Windows Mobile could grow and thrive, considering the recent market success of the iPhone. “Microsoft’s strategy for Windows Mobile has always been to nail tough business requirements while not forgetting that all of us go home to our families and friends,” Myerson said. “And we want people to carry a single phone that crosses those two worlds seamlessly.” He mentioned Microsoft’s acquisition of Danger, the maker of the popular Sidekick line of mobile devices, as an example of a move that Microsoft’s made to benefit mobile users. Given that we haven’t yet seen any Microsoft products based on the Danger acquisition, I interpret that to mean that we’ll see some Danger-based stuff in the forthcoming Windows Mobile 7 release.
Finally, I asked about “co-opetition,” the practice of both cooperating with and competing against third-party vendors. Exchange has long been an example of this behavior. By licensing Exchange ActiveSync to Windows Mobile competitors, Myerson did something that was good for the Exchange bottom line but was not well-loved by the Windows Mobile team. I asked how Myerson thinks Exchange will use co-opetition in its future product strategies, especially with respect to hosted Exchange offerings—a new area for Microsoft but one where the company already has established partners. His answer was direct: “Exchange will continue to do what is right for our customers—when you are at the heart of unified communications you have an obligation to do so. This means working with competitors when appropriate and growing into new businesses like Exchange Online.” He continued by saying that “Microsoft is a partner-led company, and we will continue to rely heavily on and work closely with our partners, be it in a server or online business.”
One of Myerson’s legacies in Exchange is his reputation as someone who gets things done; he’s not afraid to make decisions and act on them. I think this is exactly what Windows Mobile needs, and I’m looking forward to seeing how Microsoft’s Mobile Communications Business (MCB) unit evolves under his leadership. What do you think Windows Mobile needs to do to thrive against its competitors? Post a comment below or drop me a line at probichaux@windowsitpro.com to let me know.
