An attendee at a recent performance tuning session wanted ideas to more flexibly react to locks on blocks than using the WITH (NOLOCK) hint....
Read the rest of entry >>
Early-bird discounted registration for the much-anticipated relaunch of the Microsoft Exchange Conference (MEC) finishes on May 18, which is tomorrow for most people. This public service message is brought to you in an attempt to help resolve the nagging question “which conference is likely to deliver the best Exchange content for me in the next year”.
The operational and technical environment that people deal with differs enormously from deployment to deployment and a single answer won’t apply in all cases. For instance, it’s pretty clear that Microsoft will use MEC as the launch vehicle for Exchange 2013 with Kevin Allison, GM of Exchange, promising that “MEC will be full of Exchange 15 content” when he keynoted at TEC in San Diego earlier this month. Therefore, if you really must learn all you possibly can about the latest and greatest version of Exchange, you’ll be one of those who packs their Mickey Mouse ears and heads to Orlando in September to join the MECfest.
On the other hand, if you’re more interested in the details of practical deployment, tips and techniques, and the nitty-gritty of current versions of Exchange, you’d probably be better off investing in a conference fee for either TEC or Exchange Connections. That is, if these conferences continue to function in a world where a lot of the available attention and attendee dollars is being vacuumed up by MEC....
Read the rest of entry >>
Windows 7 is arguably one of the most hardened and regularly updated OSes available, and now Apple and the Macintosh are in the headlines for fighting off malware and patching vulnerable software....
Read the rest of entry >>
Question: I came across a weird case recently where we had a server that was running at 100% across all CPUs for a time but when I used the DMVs to look at wait statistics I couldn’t see anything going on. How can I see what waits were occurring at the time? Answer: This is a question that came up in our Performance Tuning class this week in London after I’d lectured about wait statistics. The key point here is that in the situation described there weren’t any actual waits occurring. Many of the monitoring scripts/tools for looking at wait statistics (for instance mine at http://bit.ly/fSWeO5) aggregate all the wait statistics and return the top 95% of all waits occurring, sorted by most prevalent waits by total wait time. This is done so that your view of the pertinent waits on the server isn’t obscured by those waits that are benign and always occur (e.g. from certain system tasks like the lazywriter in the buffer pool). A side effect of this monitoring is that unless there are no other waits occurring on the system, you will not see the SOS_SCHEDULER_YIELD wait type show up, even though they may be occurring in great numbers. Why is this? The SOS_SCHEDULER_YIELD wait type occurs when a thread executing on a CPU manages to execute for its entire quantum – the uninterrupted time a thread is allowed to be using the CPU before it must allow other threads to have a turn executing. The SQL Server scheduling quanta is 4 milliseconds (and this is not configurable). When a thread exhausts its quantum, it moves off the CPU, but instead of being suspended and put on the waiter list, it moves directly to the bottom of the runnable queue in the scheduler. This makes perfect sense as the thread isn’t waiting for a resource – so has no reason to be put on the waiter list. However, a wait type must be registered and so the SOS_SCHEDULER_YIELD type is used and it will have a zero resource wait time. There may be a signal wait time though (the time the thread spends on the ...
Read the rest of entry >>
With cloud services, even Office 365, what you don’t know about your cloud service can hurt you, this SharePoint expert found....
Read the rest of entry >>
Recently I was shown a whitepaper from EMC titled "Backup and Recovery for Microsoft SQL Server Using EMC Data Domain Deduplication Storage Systems". Over all the whitepaper isn't bad giving information about how to setup EMC's Data Domain product to be used with Microsoft SQL Server so that the backups can be taken from the SQL database and written to the Data Domain VTL and deduplicated....
Read the rest of entry >>
Even in an era of massive quotas, it’s annoying to find that mailboxes are cluttered with extraneous logging messages that accumulate steadily and are never removed without manual intervention. So it is with Outlook synchronization logs, which you’ll find tucked away in the Sync Issues folder (to expose this folder, click on the Folders icon and expand the folder hierarchy). Outlook 2007 and Outlook 2010 generate synchronization logs when errors occur when clients synchronize local replica copies (held in the OST file) of cached server folders. Logically, synchronization errors and log generation only happens when Outlook is configured in cached Exchange mode. There are many reasons why a synchronization operation might experience some difficulties. Network glitches are the obvious example – something that is extremely likely to happen when connecting Outlook to Exchange Online in Office 365 when transient network errors are common between the network that the PC client runs on and Microsoft’s datacenters. After all, no one controls the Internet and no one guarantees the speed, latency, or reliability of an Internet connection. ...
Read the rest of entry >>
Concerns about cloud security have helped dampen enthusiasm for cloud computing among some IT professionals, a perception that Microsoft hopes to reverse with some findings from a study it commissioned....
Read the rest of entry >>
Few people are aware of the excellent parental control capabilities in Windows, which started in Vista and got a lot better in Windows 7. With Windows 8, Microsoft is raising the bar yet again, and as this week’s Building Windows 8 Blog post notes, they’re adding some new features too....
Read the rest of entry >>