The beginnings of "operational awareness" and end-to-end manageability
Nobody decides to pursue a career in IT because they find systems management fascinating. That's what a Microsoft Systems Management Server (SMS) MVP recently told me. And yet, IT professionals' highest priority, according to Windows IT Pro's 2006 Industry Trends survey, is "managing IT infrastructure," and their biggest pain point is "limited budgets and expanding responsibilities." Systems management may not be sexy, but it nevertheless consumes a huge amount of IT energy and effort—not to mention 70 percent of IT budgets.
And Microsoft hasn't failed to note that all the work of maintaining a functioning infrastructure not only detracts from IT's ability to innovate and deploy new technologies but also presents an opportunity for competitors to lure IT away from Windows. The company is sharply focused on the fact that its competitive advantage hinges on continuously simplifying and unifying the management experience throughout the Windows environment (i.e., Microsoft OSs and applications such as SQL Server, Exchange Server, IIS, and Office). The Dynamic Systems Initiative (DSI) is Microsoft's company-wide strategy to attack the management problem end to end, from application development to IT to end users. DSI aims to unite IT and corporate developers in creating operationally aware applications that capture IT knowledge and incorporate health models that facilitate troubleshooting and maintenance.
A year ago, I talked with Microsoft Corporate Vice President Kirill Tatarinov about DSI and his Windows Enterprise Management Division's (WEMD's) System Center products, which were being developed with the vision of bringing DSI to life by enabling self-managing dynamic systems ("Radically Simplify IT," April 2006, InstantDoc ID 49503.) This year, as the latest version of System Center products begin to reach the market, I again spoke with Tatarinov, as well as System Center General Manager of Marketing Larry Orecklin, to discuss the products and how they address your priorities and pain points.
Forster:What would you say are the two most important DSI developments in the past year?
Tatarinov: We worked closely with the industry to take their feedback and fine-tune the strategy and initiative. A couple things happened: One is standards-related. DSI revolves around the very important concept of applications that are "designed for operations." But we have a robust hardware partner ecosystem, including networking vendors, storage vendors, and ISVs. Unless we have a standard that enables people to define their systems using a language that all the partners can understand, we won't be able to fulfill the "designed for operations" dream. One of the biggest realizations we had about DSI is that the language for expressing system constraints and the meta-model needs to be standard. That was the driver for turning our proprietary SDM \[System Definition Model\] into the published specification called SML \[Service Modeling Language\].
Orecklin: SML is how you describe an IT service, the components in that service, and the relationships between those. Customers' environments are increasingly heterogeneous. When we thought about how our SDM model compares with other initiatives in the industry, we worked with more than ten industry leaders to form the SML Working Group, which has SDM at its core. Since the initial announcement, many more are looking to join.
Tatarinov: IBM, Cisco, EMC, HP, and others are helping take the original specification to the next level, and in the next three to four months, hopefully, make it the industry standard.
Orecklin: There's also an industry initiative called the Configuration Management Database \[CMDB\] Federation Consortium. CMDB helps define and catalog all \[IT\] assets and components and the state of those assets. We joined that group and are working with them to adopt SML as the core language and modeling infrastructure.
From a designed-for-operations perspective, standards is the key technical movement. All our products are leveraging this common model infrastructure as a way to describe and capture the knowledge that exists all the way from the developer through to the end users. It's incorporated in Visual Studio \[VS\}. It's a core part of the System Center portfolio.
Forster: You mentioned two key results that came from industry input. The first was the standards you just discussed. What was the second?
Tatarinov: The second is fine-tuning DSI by considering a new persona: the business architect. This concept originates with feedback we got from industry analysts. This persona thinks about the connection of business and IT. So now DSI addresses the developer, IT professional, and business architect to provide the CIO an enterprise governance view.
How do you define the ecosystem and the collection of tools that will plug in together and deliver complete umbrella-style management so that the CIO would be able to understand and see a concrete set of reports that span project management, asset allocation, governance and compliance, traditional IT infrastructure management, and development aspects all coming together? The business architect persona fulfills the CIO's dream in that scenario.
Forster: What's the purpose of these personas in relation to DSI?
Tatarinov: DSI works by connecting several products that fulfill an individual persona's needs. The VS brand is for developers and architects. System Center is for IT managers. Microsoft Project and Microsoft Office are for business architects. The connection happens through standard interfaces, standard schemas and models, and point-to-point connectors that are being built.
We also define very crisp scenarios for how those connections work. A simple scenario: I'm the developer. I built the system. The system automatically gets provisioned and goes into operations. When operations sees an alert, that alert is automatically mapped back to the developer environment and gets logged as a bug for the developer who built this system. Then the bug can be corrected, and the fix automatically finds its way back into production.
Forster: How does DSI differentiate System Center products from third-party products?
Tatarinov: All the System Center products share DSI-based characteristics: First is ease of use (and I put ease of deployment in the same category). Management products have been hard to use and learn and require significant consulting engagements before they can be deployed and scaled. This is something we've tried to reduce.
Second, System Center products are driven by knowledge that we assembled from the industry and from focusing on our customers. For example, we worked to understand the backup and restore needs of our SQL Server, Exchange Server, and SharePoint customer base. We also spent time with Exchange customers to understand what they need to proactively correct errors with as little downtime and manual intervention as necessary. The knowledge we gained has manifested in the System Center products and is a critical attribute and differentiator for our products.
Third is scalability. We scale up to the largest enterprises out there and down to the smallest organizations. Scalability up and down is an important differentiator.
Forster: System Center Operations Manager 2007 (Ops Manager), the successor to Microsoft Operations Manager (MOM), is the first product to be renamed under the new brand. How does it incorporate DSI?
Tatarinov: With System Center Operations Manager coming out in April, customers can take advantage of the model-based management approach and apply it to management of services. You will be able to define the model of a service and manage that service the same way end users see that service.
Orecklin: Historically, the industry focused on monitoring things—the server, the application. But it's crucial to look end-to-end at how to deploy, manage, and monitor a service. Take messaging, for example: I care about my Exchange server, the network, my SAN, and AD. I need to ensure that each component is being managed and monitored but also that they all roll up to provide an end-to-end view of the entire messaging service. To do service-level monitoring, \[Ops Manager has\] new views, templates, and wizards to easily define and create a template out of the box. That includes management packs for all Microsoft components, as well as for our partners' on hardware, network, storage, and so on.
Next, you can capture knowledge about the desired state and health of an application and easily monitor and manage that over time. Inline tasks are right in the UI when things go wrong, and all the knowledge base associated with that application is right there. We've also extended the concept of knowledge down to the client. When errors occur, we can capture knowledge about an application, OS, or hardware from clients. You can view client information at the enterprise level or at the group level and link to the Microsoft Knowledge Base. You don't have to wait for a user to call about a problem on a machine.
The least intrusive level of client management is agentless exception monitoring. The applications surface up their events so the customer can monitor and report trends. We also have client management packs for Windows Vista and Office 2007 so we can actively monitor and manage business-critical clients.
Forster: System Center Configuration Manager 2007 (SCCM), the next release, is currently in private beta and goes to public beta in early Q2 of 2007. RTM is set for summer or early autumn of 2007. How does SCCM fit in the DSI picture?
Tatarinov: Customers get the ability to use SCCM as their definition and enforcement mechanism to apply a model-based approach to defining the desired state of their environment and making sure that environment stays consistent with their desire.
Orecklin: Desired-configuration management (DCM) is where the notion of knowledge and models comes into play. You can use modeling to define the desired state of a client and an application and then monitor that over time to identify drift from the desired state. From a security and compliance perspective you need to monitor and manage drift and either automatically update or take an action.
Obviously, this is a big year with Vista, Office, and Longhorn Server coming out, and companies are looking for help. SCCM's first focus is radically simplifying OS deployment. In the past, this has been a complex and manually intensive process. We provide a single integrated tool for each OS image "instance." For example, SCCM provides an integrated view of desktops, laptops, and servers. We found that many customers are maintaining dozens or hundreds of OS images because of varying hardware driver sets. SCCM provides a Driver Library so that IT can decouple the drivers from the core image, significantly reducing the number of OS images necessary to maintain their user base. In addition, preparing an OS image for deployment revolves around dozens of individual tasks such as configuring security settings, joining domains, and so on. So we developed a new feature called the Task Sequencer in which dozens of tasks are available and an administrator can drag and drop tasks in the correct sequence for each user set. Finally, customers asked us to improve ease of use and reduce complexity. For example, deploying a patch in an enterprise with SMS could take as many as 18 screens. With SCCM, that's down to as few as six mouse clicks.
System Center Essentials
Forster: Readers in small and medium-sized organizations tell me they are excited about System Center Essentials (SCE), which is slated to ship in the first quarter of 2007.
Tatarinov: SCE plays into a market where customers have fewer than 500 PCs. This market segment has been underserved historically. People are starving for a solution. We talked about different personas. This person has to deal with every aspect of IT, and we want to enable that person to do everything with a simple, easy-to-use interface. SCE will enable that person to easily configure devices—be it servers or desktops—and distribute software to those devices. SCE also lets this person monitor network, servers, and devices in a very simple way.
Virtual Machine Manager
Forster: Virtualization is the hottest technology in our industry, and competition is already out there. System Center Virtual Machine Manager (VMM) is in beta, scheduled for launch in Q3 of 2007. What's your competitive angle on managing a virtual environment?
Tatarinov: We think you can't look at virtualization in isolation. Our approach to virtualization management is to bring it under the context of the infrastructure and enterprise management overall. System Center Virtual Machine Manager is the product that will extend Operations Manager and Configuration Manager into the domain of virtual machines and enable those products to provide seamless management of both the physical and virtual environment. This is the core differentiator for Microsoft compared to other players in that space.
Forster: What are the challenges in managing a virtual environment?
Tatarinov: It's a whole lot more dynamic compared to physical machines. Things like rapid discovery and capacity-based and on-demand provisioning become much more important than in the physical world and are done on a much more frequent basis.
Forster: How do you differentiate VMM from competitors such as VMWare?
Tatarinov: We're combining the management of physical and virtual environments, and we enable people to use the same interfaces to manage their entire application, entire service, whether it's implemented on a physical or virtual machine.
In Longhorn Server, with Windows hypervisor, we're now thinking of virtualization as a component, or feature, of Windows as opposed to being something standalone. A big differentiator that customers recognize is that Windows has virtualization as a feature.
Data Protection Manager
Forster: System Center Data Protection Manager (DPM) has been a successful disk-based backup solution for file and print servers. The demand for DPM to also back up SQL Server and Exchange has been high since DPM launched.
Tatarinov: DPM V2 will support SQL Server, Exchange, and SharePoint. DPM V2 also provides archiving capabilities and works directly with tape drives. We're enhancing and simplifying the UI, which is already much simpler than what the rest of the industry could offer. It's going to be a killer product.
Service Desk and VSTS
Forster: The new System Center product code named "Service Desk" is currently in private beta, with a public beta slated for April 2007and RTM for a year later. What is Service Desk?
Tatarinov: The product provides a platform for end-to-end IT management and a framework to build solutions on top of that. Service Desk includes a workflow engine that will provide the basis for how we automate IT processes, and the implementation of the SML-based CMDB, which will be the foundation of our asset- and change-management capability. Following DSI's principle of capturing knowledge in models, Service Desk will include workflow templates for key customer scenarios. Service Desk will also deliver unprecedented integration with both Operations Manager and SCCM.
A very important platform aspect of Service Desk is a self-service portal. We're focused on enabling end users to do as many things as possible. So IT pros can define a policy. Then that policy is applied to the organization, and the end user is empowered to automatically do things that the policy allows.
Forster: Service Desk seems to bring "designed for operations" full circle by providing a means to feed production and user data back into the development process through VSTS (Visual Studio Team System). (Sam Guckenheimer, a group product planner in the VSTS group, explains the role of VSTS in DSI in the Web-exclusive sidebar "System Center ‘Service Desk' and VSTS: Where IT and Dev Meet," InstantDoc ID 95147)
Tatarinov: "Designed for operations" is a prime DSI concept—DSI being the connector of the entire system life cycle. Manageability and operational disciplines need to come early in the cycle, and everybody who builds the system needs to think about manageability. They need to be creating health, configuration, and task models early in the design phase rather than employing the traditional approach, which was: Build the system first, it goes into production as a black box, and then someone else—like a traditional systems management vendor—comes in and pokes at that black box to find out what's going on. You can't manage the unmanageable. If a system is created as an unmanageable black box, it's going to remain a black box and you'll just spend more money trying to manage it. "Designed for operations" means there are no black boxes. The system is created to be easily put into production and easily managed.
Forster: Integration of IT knowledge is a core tenet of DSI, so how does Service Desk incorporate that knowledge?
Tatarinov: The knowledge we assembled and put into the market in the form of Solution Accelerators will be encoded in Service Desk. Another important aspect: Every serious IT organization has little books in which its knowledge is written. Those organizations will be able to encode that knowledge and make it residual. In Service Desk, you'll be able to define best practices and policies for applying change or managing assets and for levels of approval, and it's all going to live in the product.
Doing the Right Thing
System Center isn't going to make systems management sexy, but the vision of enabling self-managing dynamic systems is going to help IT deal with its highest priorities and greatest pain points. By focusing on simplifying IT, Microsoft is protecting and conserving its greatest asset—its customers.