Jeff Jaffe’s Blog

In a previous blogs, I described Novell’s leadership in open source virtualization. We were the first Linux distribution to introduce XEN and have significant customer experience through Open Enterprise Server (Netware virtualized on SUSE Linux Enterprise Server). We are at the forefront.

I have also emphasized our focus on interoperability – fundamental to our brand promise of Making IT Work as One. Our products play several roles in a data center based on customer need. So our virtualization strategy is not only deep – it is also broad. We work with partners to give customers outstanding results – no matter what their existing IT footprint looks like, and no matter what role Linux plays in their enterprise.

Several recent partnerships punctuate this point:

Microsoft’s Server Virtualization Validation Program (SVVP)

Last month we joined the Server Virtualization Validation Program (SVVP), Microsoft’s program to validate Windows Server 2008 running as a guest on a different system. We’re busy working to validate Windows Server 2008 on SLES.

This is a continuation of our virtualization partnership with Microsoft. Nearly two years ago we agreed to optimize SLES on Windows; and Windows on SLES. We built an interoperability laboratory to ensure early and constant focus on interoperability, optimization, and support. SVVP is the next step: we’re workijng with Microsoft to construct the program and are now working on validation.

VMWare’s Virtual Machine Interface (VMI)

Sometimes Linux is the virtualization host, and sometimes it is the guest. Today’s market leader in virtualization is VMWare, and so if customers run a Linux guest over VMWare we ensure they get outstanding performance using SLES.

Several months ago, VMWare announced VMI to improve the performance of guest operating systems over VMWare. We immediately built those features into SLES. Last month we pointed out that SLES 10 SP 2 incorporated VMI to further build our VMWare relationship.

PlateSpin

When I described Novell’s motivation in acquiring PlateSpin, I mentioned that PlateSpin’s commitment to interoperability was critical. PlateSpin is agnostic to platform, supporting many operating systems and virtualization platforms. We have been clear: PlateSpin will continue that approach.

Last month, with the introduction of PlateSpin PowerConvert 7.0 we took it to the next level: We enhanced PowerConvert support for Linux including Red Hat Enterprise Linux. (Sure, we included SLES, as well – but the point of this posting is interop!) We also provide Windows 64 bit support for physical-to-virtual migration, both for VMWare and Citrix’s XEN Enterprise solution.

Truly a demonstration of our commitment to play multiple roles in the data center.

Management of Linux and Microsoft’s System Center

As I’ve mentioned, part of our Linux strategy is to provide enterprise quality; with key, mission-critical properties such as reliability, manageability, security, and interoperability together with unparalleled support. To provide a complete solution, we provide both the core Linux distribution and companion products such as ZenWorks Linux Management. One vendor – with a complete industrial strength solution.

But, we recognize the realities of a heterogeneous world. So Linux has to be well managed in customer environments that have other management systems.

We were pleased to announce in April this year that we are building an Advanced Management Pack which will give Microsoft’s System Center the ability to be a fine manager of SLES. SLES is the Linux of choice in an environment managed by Microsoft’s products.

In doing so, we achieved another breakthrough in our partnership. Microsoft publicly committed to provide open source code to collect management data and make data accessible to management services. There is a growing recognition of the role that open source plays; both in creating technology – and also for standards.

Novell continues its commitments to interoperability; adding many customers scenarios. Keep watching this space for more!

In my last posting I described initiatives to ensure excellence in our internal processes. I described Integrated Product Development (IPD) which we use to gain information about customer needs, market trends, and technology trends and translate into high quality products that meet these needs and wants.

While IPD is used to decide what products we build, the processes which assure that they are built appropriately – on time, with quality and efficiently – are our Engineering Excellence processes. Since Novell business units came from Novell and acquired companies, we have a collection of engineering processes to choose from.

Waterfall and Scrum

There are different approaches to building software. Traditional waterfall techniques go through structured phases: architecture, design, development, unit test, system test, integration test, and regression are all typical phases. In agile techniques there is continuous build of what has already been developed, and all code is continually integrated and tested to create a potentially releasable product. With our focus on open source, we are involved in community approaches; participants are involved inside and outside our company; and we maintain both a community and enterprise version of the product. There are other variations: spiral, scrum, and extreme programming are popular.

In looking at these choices, we determined that we would get good coverage if we focused our energies on two typical approaches: traditional (waterfall, modified waterfall, the V-Model) and scrum. By standardizing on these two, we largely cover all of the choices – including open source development.

What are some key engineering processes?

There are many Engineering processes. Included in them are:

• Interlock with product management on feature selection
• Project/release planning
• Project management
• Estimating effort
• Putting in the proper buffers to deal with expected but unforeseen delays (e.g. requirements churn)
• Globalization (e.g. double-byte)
• Localization (of content and interfaces)
• ISV support
• User-centered design
• Backlog management (especially for scrum)
• Criteria to exit from one phase to another of development
• Product ship criteria
• Quality metrics and targets
• Testing strategy
• Beta criteria
• Design documentation
• Patch and update
• Multi-site development
• Defect management
• Coding and code inspection
• Documentation
• Tools (e.g. configuration management, bug detection)
• Vendor management
• Re-use
• Incorporation of open source assets
• Security
• Export
• Architecture reviews
• Install
• Common technologies (e.g. user interfaces, databases, directories)

A phased approach

Novell has an excellent reputation for its engineering prowess. For this reason we only recently have selected Standard Engineering Practices (SEPs) in our Engineering Excellence initiative.

We did not rigorously formalize all of the above processes at once. While standardization has benefits, we are careful not to overindulge in process, since that can slow down development in a way that is counterproductive. This year, we have chosen a small number of SEPs to get started. These include the ones that are most critical to developing products in a timely fashion and with high quality. These SEPs relate to: interlock with product management, backlog management, project/release planning, estimation, buffering, acceptance/exit criteria, ship criteria, and design documentation. Even these are not brand new – the standardized processes rely on tried and true processes that parts of Novell have been using for years. And it is our intention to add new SEPs every year -as part of a continuous improvement program.