After the style, some substance. Having dished out the Windows 7 eye-candy for software developers Microsoft's dipped into virtualization and system scalability for the hardware heads.
Executives have promised Microsoft's Hyper-V virtualization technology, mauled mercilessly before launch to hit deadlines, will deliver on pre-launch promises.
Bill Laing, corporate vice president of Microsoft's Windows server and solutions division, told The Reg planned management features would be the competitive differentiator in the next version of Hyper-V, due with Windows Server 2008 R2.
Laing was speaking during Microsoft's Windows Hardware Engineering Conference (WinHEC) in Los Angeles, California, where the company had demonstrated Hyper-V performing a live migration, playing a video while switching hypervisors.
Hanging in the air was Kilimanjaro, the next version of Microsoft's SQL Server database that promises greater scalability and improved management in large environments than the current SQL Server 2008.
The SQL Server RDBMS engine will be changed in Kilimanjaro to run up to 256 logical processors, going beyond the limit of 64. Quentin Clark, general manager for the SQL Server database engine told The Reg Kilimanjaro will eliminate the need to manually partition applications across nodes, allowing SQL Server customers to run "very high-scale and difficult-to-partition applications."
Microsoft is "lighting up" new technologies with Kilimanjaro that will have a greater focus on business intelligence, he said. For example, Excel will be able to work with larger data sets than currently possible. "We are aligning it with more Office experience...to complete the last mile of business intelligence," Clark said.
There's also the potential for reference implementations of SQL Server tailored to more than just business intelligence.
Microsoft last month announced Madison, a server-based appliance based on Kilimanjaro with hardware partners for large scale and detailed analysis.
Microsoft plans reference implementations ahead of Madison and said it may or may not expand into other areas as there'd already been early interest. This could mean appliances for packaged applications, like SAP, or - much harder - "custom old transaction processing."
The latter could be harder to achieve because reference architectures tend to require a knowledge of I/O, memory and CPU capabilities in addition to knowing a thing or two about the software's own capabilities. That's relatively easy in packaged apps, less so when it's the user's own software. ®