First of all, the delay on this post must be because I'm too techsexy.
The first event Tuesday morning was the kick-off keynote by Bob Muglia, Senior Vice President, Server and Tools Business. Actually, the keynote session first opened with Rodney Sherwood, claiming to be the Senior Vice President, Humor and Comedy Services. (A few alumni I spoke with later said he's the regular crowd-warmer for Microsoft events.) Bob's keynote focused on the Dynamic Systems Initiative (DSI) that is a ten-year plan started in 2003. DSI ties in with the Infrastructure Optimization model: basic, standardized, rationalized and dynamic. The goal is to move businesses through each stage towards a truly dynamic system.
Bob talked on three pillars: virtualized infrastructure, design for operations (using service modeling language [SML]) and knowledge-driven management. Jeff Woolsey, Lead Program Manager, Virtualization, did a demo of a single server running Longhorn Server Core (a stripped-down, no-GUI version of Longhorn that works well as a base for virtualization), running Longhorn server host and then four guest OSes: 2003 32, 2003 64 w/ SQL 2005 64, Suse Linux, and 8-core Longhorn 64. He then demo'd Virtual Machine Manager and Operations Manager, which integrated together with a built-in PowerShell script to hot-add a NIC to the SQL server. It also has hot-add capabilities for memory and processors.
System Center Operations Manager use SML to store system and application models. Microsoft is now partnering with EMC and Cisco for foundational model creation and plans to create industry standard models that can be used by any organization, or customized to better match their systems and applications. SCOM should RTM in early April. Barry Shilmover, Program Manager, Management Practices, demo'd SCOM. He showed that it is based upon models so can provide service-level management not just hardware monitoring. The next release of SCOM will have network-aware service management through a partnership with EMC. A bi-directional connector between SCOM and EMC Smarts provides some of this now; it will be be built-in (and appropriately licensed) in future SCOM releases.
After the keynote I attended System Center Configuration Manager 2007: State of the Union presented by Bill Anderson, Lead Program Manager, SCCM, and Brian Kauer, Group Program Manager, SCCM. It was a VERY popular session; standing room only. They first introduced one of the marketing plans for SCCM 2007: TechSexy. They used a series of images to really explain the concept: fanny pack (not techsexy) versus new hi-tech clothing (internal gadget pockets); Neo from the Matrix is techsexy; Tron is techsexy. There was some debate as to whether this is techsexy. They announced that HP is releasing a Client Catalog for SMS which will do custom updates for HP client systems (most everything buy BIOS) making it the largest custom update package. They did a quick show of Asset Intelligence (AssetMetrix) in SMS SP3 which adds more than thirty reports with data like software serial numbers, product IDs, and purchase channel (e.g., OEM, volume, etc.).
On SCCM 2007 they said that operating system deployment was the biggest investment. A company (I missed the name) just announced desired configuration management packs for industry compliance (SOX, HIPAA, etc.). A new server role, Fallback Status Point (FSP) logs problems with clients that can't communicate with the management point. In the console, boundaries are now a node under site settings (instead of buried in the site properties), collection columns are properly sortable and customizeable (add things like Last Update Time), Search Folders are available in many nodes, and actions like multi-select and drag-and-drop are now available. Another marketing campaign is Designed for BIG.
The next session I attended was SMS 2003 SP3: Asset Intelligence presented by Quincy Milton, Group Manager, SCCM, and Michael Nappi, Product Unit Manager, Asset & License Management. AI with SP3 is an optional install as it extends both the database and MOF. There are 30+ new reports in three major categories: hardware, licenses, and software. A few examples: Computer age (estimated by CPU age), Primary user (user logon count), hardware not upgrade ready, and software families. Many reports utilize Vista's Software Licensing services, a basic version exists in XP. SP3 RTM target is May 2007.
SCCM 2007 will have all of the AI features from SMS 2003 SP3 plus change management summary, asset tagging via OSD and CAL utilization reports. System Center Online Services is a hosted service of a live knowledge base (AssetMetrix IP) to keep the software catalog of 350,000 titles up-to-date. It will eventually have feeds from ISVs, OEMs, partners, customers, etc. SCCM will have a bi-directional connector with Online Services for synchronization. This same model will be applied elsewhere in the System Center family.
After lunch I went to System Center Configuration Manager 2007: What's new in Software Distribution led by Dave Randall, Program Manager, Software Distribution. He first went over some general new features of the console. Organization folders, which existed in SMS 2003, now fully replicate to child sites. Search folders are technically the same as organization folders. Most nodes have a home page that displays useful data, but it's not live, just recent. The top-level home page will load the TechNet - SCCM page (if available) allowing users to see the latest news. Home pages have resources and Internet links, e.g., checklists. Administration Pack Tools are now integrated. The RC version will replicate the organization folders under Packages and Advertisements to the similar folders under the Status node. Node filtering is available: at the top of the list you can enter a value to "look for" in "all columns" etc. Each node also now shows the count of items.
Each maintenance window has a maximum of 24 hours, but multiples can be applied to a single collection to form a larger maintenance window. When upgrading to SCCM 2007 the prerequisite check will flag all "unknown" program runtimes. Gotcha: last five minutes of the maintenance window is a restart buffer (e.g., something assigned in those last five minutes won't occur until the start of the next maintenance window). Maintenance windows can be added to a collection but disabled if not in use. There are a few reports available (but use the phrasing "Service Windows"), e.g., Service windows available to a particular client.
A Branch Distribution Point (BDP) can replace a secondary site at low bandwidth sites as it uses BITS to transfer from the standard DP. BITS throttling is now centrally configurable. Other site systems can be protected, not just DPs. If the target system is not an installed client or not assigned to the site it cannot be enabled as a BDP. The Copy Package Wizard will display any organization or search folders and can copy from a specific source for DP cloning. Wake-On-LAN (WOL) packets are sent serially from the site server, three packets per target client. Advertisement program re-run options: never, always or if previous failure/success. Custom policy polling interval can be set on collections; conflicts with multiple collections are resolved by shortest time. Branding support in the new Computer Client Agent; can display custom text to the client (including URLs).
My last session of the day was SCCM Deployment, Part 1, presented by Dave Randall and Prabhu Padhi, Program Manager, SCCM. During SCCM installation detailed setup progress notification is provided so you know exactly what's happening while waiting. One new role is the State Migration Point, which is like a reverse DP (client pushes up USMT data). Two recommended site systems are the Fallback Status Point (FSP) for client troubleshooting and the Branch DP for remote sites. The Management Point (MP) and Software Update Point (SUP) site system roles should be separated, at least during initial testing, for easier troubleshooting. If using 64bit hardware the Reporting Point (RP) must be separate (or virtualized) because IIS must be in 32bit mode.
CCMSetup is now a single client binary. There are multiple Client Deployment reports available if the FSP is used (specified in the client install command line or registry). During installation the client downloads a manifest and then only downloads the required components (instead of downloading all of the components and then evaluating the installation requirements). Client installation options can be written into AD so the client installation command is just ccmsetup. CCMSetup.msi can be used as a Software Installation package so if it is removed from AD the client is not uninstalled just the CCMSetup wrapper. Another preferred (and techsexy) option is to publish the CCM Agent to WSUS as a required infrastructure update (similar to the WU Agent).
A tip for deployment is to gradually phase in the clients, e.g., 5-10 at first. Then do a "policy loop" check sending small advertisements to check all settings and functionality. Tweak the site settings (i.e., faster cycles) initially during ramp-up to accelerate data flowing into the site, and then scale back after deployment.
Clients send new state messages to the MP along with the regular status messages. State messages provide more progress indication rather than just start/end status. Software update reporting no longer uses hardware inventory but uses state messages.
The client requirement for .NET Framework is really only for Desired Configuration Management (DCM) so it's not a hard requirement.
OSD and Device Management packages are transferred in an upgrade and can be used as legacy packages, but need to be retooled for full functionality. Top-down upgrades are required possibly even installing a new SCCM central site above the existing parent. Standard security won't upgrade, legacy clients won't upgrade and there is no SMS 2.0 interoperability or upgrade available. SCCM mixed mode allows for backward compatibility with SMS 2003.
SCCM backup uses Volume ShadowCopy Service (VSS) so the site comes back up faster while the shadow data is archived in the background.
Site systems can be configured to either push or pull with other site systems. Aside from basic status messages, site servers send additional heartbeat status for availability monitoring.