Jeff Jaffe’s Blog

Raison d’etre

On November 2, 2006 Microsoft and Novell announced a landmark collaboration focused on Windows / Linux interoperability. This has been successful from many points of view. Our business collaboration got off to a quick start as illustrated by press releases over the last several months where customers have bought in.

But the real promise of the collaboration is in the technical collaboration. Here is where our two companies work together to address customer problems in interoperability. With Linux and Windows being the two strategic platforms for most I/T managers, making these environments work together is job #1.

Earlier this month Microsoft and Novell provided our technical roadmap for interoperability. The technical work on our interoperability projects has been yielding results almost immediately (see our December office interoperability press release.) But we knew that specifying precise commitments would take time, so we agreed in November to take several months before mapping out our roadmap. This was achieved earlier this month.

Different types of interoperability

The first challenge was to decide where to focus. To make two disparate systems work together involves every layer of architecture. It involves APIs, protocols, and languages. It involves server issues and client issues. It is addressed with intersystem translators and intrasystem API mappings. It includes management, directories, security, virtualization, documents, web protocols, media players, codecs, print servers, file systems, databases, graphics, and animation. Where does one begin and how to make progress against such a wide diversity of candidate areas?

We found it useful to decompose the problem into prioritized and manageable pieces. This is the heart of the technology collaboration roadmap announcement last week. The decomposition is side-by-side interoperability, bottom-top interoperability, and management.

• Side-by-side interoperability: Side-by-side is the common environment that customers have today. They have both Windows and Linux clients. They need to share data. They have Windows and Linux clients and servers. All of these need to be members of a common I/T infrastructure.

To address the side-by-side interoperability problem in all of its glory indeed requires addressing all of the problems referenced in the first paragraph of this section. But then how did decomposition help? We are back to solving the universal translation problem!

Project 1: Documents. We made the following observation. At the pure client level—the most critical day-to-day need was to be able to share documents. I can remember meetings of the Computer Systems Policy Project (CSPP) in the mid-1990s in Washington, DC. The problem of document interoperability was primary then—and it continues to be the most important aspect of sharing. So we chose this as one of the four focus areas to get started with.

Within Novell we experience this every day. We are a Linux shop. We use SUSE Linux Enterprise Desktop (SLED) which utilizes the Open Document Format (ODF) Standard. With SLED 10 we have made enormous progress in terms of interoperability with Microsoft Word. But it is not 100%. So it is easy for us to appreciate the importance of document interoperability.

From a standards perspective, Novell has always been a supporter of ODF. But Microsoft Office is a reality, and the marketplace simply demands that we interoperate with it.

Project 2: directories and identity management. While document interoperability is the most immediate client interoperability problem, there is also a need to address broader issues. This is more complex, because, as mentioned earlier, there is a plethora of potential issues to address. All of them solve some aspect of interoperability—none solves everything.

We determined that the most immediate area was directory interoperability. With clients from a variety of networks being able to register and gain access to services on multiple networks, we would have a key first step in driving systems closer together. True, for complete interoperability, other support would be required later. But at least we have taken a first step.

We’re now taking it a step further. If we are getting networks to interoperate—why stop at Windows and Linux? After all, our own NetWare customers have for years wanted to have improved interoperability with Windows environments. So we’ve enlarged the scope further. We are considering environments supported by Active Directory, eDirectory, general LDAP directories, and new open source identity frameworks such as the Bandit project.

• Bottom-top interoperability: virtualization. The introduction of virtualization technology provides new opportunities for interoperability.

In this scenario, there is not a need at an end user level to be interoperating between two environments. Rather, there is a need at an I/T infrastructure level to support multiple environments on a single physical server. This is achieved with virtualization. With virtualization, there is a single host machine (typically Windows or Linux, although in principle it can be anything), and the host supports multiple guest operating systems running on it.

Why do I/T managers require this type of interoperability? The most common reason is that they are consolidating function onto a single server. This could be to reduce hardware costs, reduce physical footprints, or simplify management. Several other reasons are mentioned in an earlier post. In the multi-OS environment, a particular reason is to add new applications. One might have created an all Windows shop or an all Linux shop from a management perspective. In that environment, the I/T manager has decided that the host OS for all physical machines must be uniform. Nonetheless, if there is a desirable application that runs only on a different OS, the I/T manager can install the application by running it on a virtual guest.

• Heterogeneous management. The other key interoperability problem is systems management. The area of management is an area where customers demand uniformity. They don’t want to have vastly different management systems for different operating environments. So it was natural to focus on this as the fourth area of interoperability.

Summary and next steps

In this posting I have explained how we selected specific interoperability areas from a myriad of possibilities. We focused on bottom-top interoperability and management; and then picked off two of the key projects in side-by-side interoperability: documents and identity management. In the next posting, I will describe the actual roadmap in detail.

Over the last several months, I have posted about virtualization technologies. In characterizing Suse Linux Enterprise Server (SLES) 10, I mentioned that first-to-market with XEN virtualization was a key differentiator. I talked about Infoworld’s virtualization symposium, and how the whole world was going virtual. Later in the fall, I touted virtualization (SLES on Windows and Windows on SLES) as a key value element in our agreement with Microsoft. Last month, I illustrated how virtualization was a key element providing interoperability – hence critical to Novell’s mixed source strategy. Finally, in my last major post, I illustrated the power of virtualization in addressing Netware to Linux migration in Open Enterprise Server 2.

In all of this, something has been missing. How do we manage this powerful new technology? How do you make sure you get all the value promised by virtualization? Any powerful new technology needs to be well managed to tame its power. I’ve alluded in some of these posts to our management technologies. In this post, I will outline how ZENworks is stepping up to be the solution to manage mixed virtualized environments.

My focus here is on the core capability; not the broader architectural framework. It is important however, to fit any systems management capability into a broader architecture. Please look here to find Novell’s blueprint for the general problem of ITIL (Information Technology Infrastructure Library) services based management. The virtualization management functions that I describe below are a piece of that blueprint.

Review of definitions

Let’s first review some definitions. With virtualization, one may run several guest operating systems over a hypervisor that provides some basic division of system resources. Some examples include VMWare, Microsoft’s Viridian, and the Xen Open Source project. A hypervisor might be a packaged as a standalone piece of code, or it might be embedded within an operating system. Thus Novell’s SLES 10 product has the XEN code embedded as part of the operating system. In that case we refer to SLES (or Linux) as the host operating system for this environment.

Operating systems that run on top of these hosts are called guest operating systems. These could be copies of the same code as the host (e.g. SLES on SLES), or could be different operating systems.

Virtualization is used for various reasons. One reason is to optimize performance of compute resources. This is achieved by running multiple applications on the same computer. Another reason is to provide some isolation between applications (by running them in separate virtual machines). Still another reason is to reduce complexity by consolidating servers and having fewer total servers to run; reducing cost and footprint. Some use virtualization to aid with high availability. Lastly, virtualization is popular as a means to create multiple images; e.g. in a remote testing environment or staging the rollout of new function to production.

Moreover, in the future, virtualization becomes the platform for services oriented I/T (look here for more on this).

Virtualization environments that are popular today

The market has evolved different approaches to virtualization. Novell has always been in the forefront of supporting all of the approaches.

VMWare is a technology that has been around longest. Novell works closely with VMWare so that it runs well with SLES. It has the largest installed base given the length of time it has been in the marketplace.

Xen is an open source approach that has been embraced by the open source community. It is built into SLES 10. The fact that it is built deeply into SLES, together with its exploitation of hardware assists and paravirtualization, ensures that it has outstanding performance. That, together with very broad vendor support, ensures that it will have a promising future.

Viridian is a technology used by Microsoft to allow Windows to be a host operating system. It can be used as a host for Windows guests. As part of the recent agreement between Microsoft and Novell, SUSE Linux Enterprise Server will also run optimized as a guest on Windows. As this is a fundamental piece of the Microsoft / Novell technology collaboration, I will have more to say about this in a future post.

Customer needs for the management of virtualization

While virtualization continues to be a crucial technology, to fully exploit the promise requires tools that help manage utilization of virtualization.

• Customers need tools that are able to support all three popular forms of virtualization. Each form has application for different customer needs, so customers likely will run more than one virtualization technology.
• Customers need automation. Each customer’s policies might be different. So policy based automation is required.
• Customers need to manage their virtual resources in the same way that they manage physical resources. Otherwise, systems become cluttered with unused virtual resources which waste system capacity. Virtual machines need to be added and deleted.
• With numerous physical machines and numerous virtual machines available to run applications, workload schedulers that work across these environments are critical.

ZENworks as the key technology to manage virtualization.

In a previous posting, I outlined Novell’s overall techncal strategy. First, it starts with Linux, and the most disruptive operating system change in IT today. Next, it focuses on interoperability and heterogeneous systems management. Novell is the mixed source leader: introducting open source into the enterprise — most notably the data center.

With this background, it should be clear why Novell is focused on the management of virtualization. Given virtualization’s role as a way to host heterogeneous operating systems on a single system, meeting the customer needs listed above fits right into Novell’s strategy. Given the strong emphasis on virtualization in SLES, the fit is even more essential.

How is Novell addressing this? In late November, Novell unveiled a collection of ZENworks products that will address these issues. These are fundamental new offerings to help manage the data center.

First, to address heterogeneity, these products will address VMWare, Xen, and Viridian from a virtualization perspective; and Linux, UNIX, and Windows from an operating systems perspective.

The policy automation piece is addressed by the ZENworks Orchestrator product. Customers can specify the policies that are important to them and they are able to decide how to manage resources. The product also learns systems performance data to help policies optimize performance. This also can help deploy virtual machines for failure situations.

ZENworks Virtual Machine Management handles the piece that enables virtual machines to be managed in the same way that physical resources are managed: provisioning, deprovisioning, etc.

ZENworks HPC (High Performance Computing) Management is the piece that will manage distributed workloads. This works not only for conventional data centers, but handles grid management as well. Also, key to this capability is to be able to physically move data resources to processing resources.

Summary

Interoperability is taking a major step forward with virtualization. One operating systems platform can host several guests from different operating systems. Different technologies exist to achieve different virtualization capabilities. As part of ZENworks’ movement to be the key data center manager of the future, ZENworks is now poised to manage all of this and help ensure that customers get the most from virtualization.