April 9, 2004
Late last week Microsoft and Sun Microsystems announced that they have entered into a broad technology collaboration arrangement to enable their products to work better together and to settle all pending litigation between the two companies. The agreement involves payments of $700 million to Sun by Microsoft to resolve pending antitrust issues and $900 million to resolve patent issues. In addition, Sun and Microsoft have agreed to pay royalties for use of each other's technology, with Microsoft making an up-front payment of $350 million and Sun making payments when this technology is incorporated into its server products. The agreement also provides both companies with access to portions of each other's server technology to develop new server software products that integrate better. The cooperation will initially center on Windows Server and Windows Client, but will eventually include other areas such as email and DBMS. Sun and Microsoft engineers will work together to allow identity information to be easily shared between Active Directory and the Java System Identity Server. Sun has agreed to sign a license for the Windows desktop communications protocols under Microsoft's Communications Protocol Program and both will work to improve technical collaboration between Java and .NET technologies. The companies have also agreed that Microsoft may continue to provide product support for the Microsoft Java Virtual Machine that customers have deployed in Microsoft's products. In addition, the two companies announced Windows certification for Sun's Xeon servers and moved forward with the certification process for Sun's Opteron-based servers. Sun and Microsoft also agreed to a broad covenant not to sue one another with respect to all past patent infringement claims and with the potential for future extensions of this type of covenant. Lastly, the two companies are settling and terminating their lawsuit in the United States with Sun indicating that the agreements announced today satisfy the objectives the company was pursuing in the EU actions pending against Microsoft.
Admittedly, much ink has already been spilled in the trades, financial community, and general media on the kiss-and-make-up of these two egocentric computing powerhouses. While many may believe this announcement was driven by political and legal bickering, we suspect that a much more dynamic force was at play, namely customer and marketplace pressure. The individual successes that both companies enjoyed during the waning years of the 20th century was marred by a great many IT shops who deployed solutions from both vendors with high expectations but received less than desired results, especially in the areas of integration. With the budgetary realities of today, it is likely that enough customers directly and indirectly told the two, “Enough is enough, maximize our IT ROI, not your own.” At the same time, a marked consolidation has taken place in systems vendors’ ranks that has increasingly pitted IBM against HP and Sun, Microsoft against Sun, and Sun against everyone. From a purely navel-gazing perspective, self-preservation is always a good thing, and it is clear that Sun, and to a lesser extent Microsoft, needed to take action to protect options for future growth. For their customers and market overall, this is good news that may have significant impact on the ease of managing and operating IT solutions, as well as provide developers with greater flexibility in choosing technology knowing that the integration hassles of the past should be noticeable reduced.
Despite years of public bickering, Microsoft and Sun are actually more alike than they are different. Just consider that both companies are managed by driven founders, value intellectual property very highly, and represent the quintessential rags-to-riches story that lifts the hearts and minds of software and hardware geeks aplenty. Both companies take a systematic approach to solution development, with Sun adding hardware into the mix as well, and strive to integrate product offerings to create a series of off-the-shelf solutions that meet the needs of customers. In contrast, over the past few years, corporate stalwart IBM has become the great weaver of IT solutions: if a customer buys IBM hardware, IGS will integrate myriad combinations of middleware and other software into a tailored-to-fit solution. While Sun has spent considerable effort as of late to bolster its services arm, it still pales against that of IGS. The continued absence of significant market movement by HP leaves IBM as the long-term threat to Sun, and to Microsoft beyond the desktop. Today’s economic and IT realities have finally come to roost and forced these twin sons of different mothers to reconcile, drop their distracting legal shenanigans, and turn their energies to taking care of their customers, in the hopes of restoring corporate growth and shareholder confidence. With this cosmic realignment underway, both can take “peace dividend” resources and apply them with new vigor to blunt the competitive force of Big Blue.
IBM has introduced the eServer z890, a new mid-market mainframe offering based on technologies from the company’s flagship eServer z990. According to IBM, the z890 offers improvements over the previous model z800 including a nearly 100% increase in capacity of each general-purpose central processor, as well as the new zSeries Application Assist Processor (zAAP) option, which delivers a special zOS/Java execution layer designed to simplify integration of Java-based applications. The z890 also offers on/off capacity on demand (CoD) for Parallel Sysplex clustering and Java workloads, improved networking capabilities, and tools for simplifying peripheral hardware connections. The z890 is about one-third smaller than previous z800 solutions, and is available as a single model with twenty-eight different capacity levels. The eServer z890 will be available on May 28 and planned availability for the new zAAP hardware feature is June 30. In addition, availability of the next version of the zOS operating system (v1.6) is planned for September 2004. In a separate announcement, IBM marked the 40th anniversary of the System/360, the first mainframe computer, which introduced technologies including transaction processing, micro-circuitry, and databases. The System/360 is described as the largest privately-financed project ever undertaken. More than 300 patents were issued as part of its development, and over 100,000 businesspeople in 165 American cities attended meetings where the system was introduced.
At one level, IBM’s eServer z890 announcement is entirely predictable. Yet again, the company has used elements from flagship high-end solutions (in this case last year’s “T-Rex” z990 mainframe) to enhance lower end products. Yet again, IBM’s mid-market customers should reap benefits from technologies originally designed to satisfy the needs of demanding, large enterprise clients. Indeed, it is this essential predictability that has fueled the mainframe’s highly unpredictable success. For over a decade, industry pundits have consistently and inaccurately forecast the mainframe’s imminent demise. In an industry where performance (and sometimes product viability) is measured in nanoseconds, a four-decade run is remarkable in every sense of the word. Said run has not, however, been without a few rough spots. IBM’s mainframe-centric world view in the late 1980s fed many of the company’s serious problems in the early 90s. But IBM successfully changed its ways. Wisely, the company did not throw its flagship baby out with the bathwater but came to regard the mainframe for what it was; an innovation engine. IBM’s eLiza initiative (now called Autonomic Computing) was largely designed to percolate mainframe-style virtualization, flexibility, security, and automation throughout the rest of IBM’s product families. Those mainframe-derived value-adds have driven much of the company’s ongoing success.
The z890 carries on this tradition, delivering notably higher performance than its predecessor in a considerably smaller, more affordable package. While the company describes the z890 as a solution for mid-size companies, we are not sure how much demand for mainframe solutions exists in that market. However, the addition of the new zAAPs and on/off CoD for Java workloads could prove particularly interesting to non-mainframe users with a penchant for Web-based applications. These Web-friendly features are likely to be of interest to IBM’s traditional mainframe users. More importantly, they may help to lure some recalcitrant S/390 users off the fence and into the zSeries pastures. The test every vendor faces is in how to upgrade products in ways that are relevant to current clients, while also attracting new classes of customers. The proof is still some ways off, but the z890’s notable enhancements suggest that this new machine is likely to help IBM continue its four-decade-plus mainframe winning streak.
Google has announced that it will begin offering a free, Web-based email service, called Gmail. The new service would give every subscriber one gigabyte of email storage, a sizably larger offering than other free, Web-based email services such as Hotmail or Yahoo’s mail, which offer users a few megabytes of email storage. The service — which presently is in beta test with selected users -- will offer search and organizing features for email that allow related emails to be clustered together without action by users. Users will be able to search their email by keywords. The service is designed with anti-spam features as well. Google is able to offer the service at no cost to users because it will apply its search engine technology to user emails for key words that will then prompt relevant advertising to appear with the email in question.
Although the folks at Google announced Gmail on April 1st, they clearly weren’t joking about the offering. Google’s announcement has also been treated quite seriously by a number of individuals and organizations; perhaps a bit too seriously for Google’s taste. The German government has weighed in noting that searching individual emails is a violation of German laws, related to wiretaps and eavesdropping. Privacy advocates in North America and around the world have expressed concerns that Gmail could be used to correlate user information with email content, creating privacy invasions of individual users. Finally, a company in the UK has claimed that it owns the copyright to the term “Gmail” and says Google has not copyrighted the name.
Certainly the last issue noted will be the most easily remedied. Google can change the name of the service to Moogle, or Googlemail or some such, or even pay the UK concern a sum to purchase the name. But the regulatory and privacy issues are not likely to be resolved in such short order. Regardless of whether privacy concerns are raised by advocacy groups or governments, they highlight the ongoing changes as the world moves fully into an era where information becomes a currency unto itself. Which brings us to the idea of what it is that really constitutes “free.” In this case, users are not going to be asked to pay money for the service, but they are going to give up a level of privacy (not to mention the appearance of advertisements in their mail) in return for their using the service. They may not pay in the traditional sense, but they are paying, nonetheless, with information about themselves, which in many cases may be more valuable or dear than money itself to these users. While many people will undoubtedly flock to the service, we suspect an increasing number of Internet users are going to take a hard look at the idea of each and every one of their email messages being searched (even if it is done so by a machine) and wonder if that is indeed worth the price getting something for “free.”
The Storage Networking Industry Association (SNIA) announced that the first Storage Management Initiative Specification (SMI-S) conformant products have passed the SNIA Conformance Testing Program (SNIA-CTP). According to the SNIA, this marks the first time that customers can purchase products built using a tested and standardized management interface which will aid in the deployment and management of multi-vendor storage environments. Overall, fourteen companies submitted more than 100 hardware and software products that passed the SNIA-CTP. These companies include Brocade, CNT, Dell, EMC, HDS, HP, IBM, LSI Logic, McDATA, Network Appliance, QLogic Corporation, SGI, StorageTek, and Sun.
Managing heterogeneous storage infrastructures has been one of the great IT bugaboos of the past half decade, but not without reason. The fact is that storage customers, like any customers, are sometimes value-conscious to the point of idiocy, the result being datacenters crowded with bargain-bin solutions that offer little in the way of integration or interoperability. Storage vendors themselves contribute to this difficult situation, though again not without reason. Market pressures, especially those of the recent, savage downturn, tend to keep competitors at a distance rather than to inspire bosom buddy cooperation. Forward-thinking storage vendors such as EMC, HDS, HP, and IBM have announced various cooperative agreements and API exchanges, but those activities have largely occurred during the past year or so, as the SNIA was working up to these first SMI-S conformance tests. While the successful completion of those tests might appear to eliminate the need for such private API agreements, the opposite is actually true.
Effectively managing heterogeneous storage environments depends on temporal as well as practical storage issues. As a result, leveraging SMI-S compliance will depend on both the range and vintage of a vendor’s solutions. The fact is that most storage environments tend to be accumulated incrementally over a number of years. While some vendors have embraced SMI-S with enthusiasm, extending compatibility backward across multiple generations of storage arrays and applications, others are pursuing compliance more conservatively. In such cases, API exchanges and similar agreements could mark the difference between a wholesale improvement in how a company manages its and entirely new species of familiar old headaches. While we applaud SNIA and its members for their efforts in promoting and delivering interoperable solutions, storage customers would be wise to temper their enthusiasm for SMI-S compliance with a caveat emptor reality check. Until end users thoroughly understand which storage solutions they have deployed, which are and are not SMI-S compliant, and which may be affected by extraneous vendor agreements, centrally managing heterogeneous storage environments is likely to remain a complex, even elusive pursuit.
This week in conjunction with its celebration of forty years of mainframe, IBM announced the ESS Model 750, a midsized version of its Enterprise Storage Server family. The new server scales from 1.1TB to 4.6TB, and is focused on price/performance metrics to provide high-end functionality at midrange prices. The new ESS Model 750 has redundant parts to support mission-critical levels of availability, and for business continuity has advanced copy services functions and management tools. For companies whose needs scale greater than the size of the ESS Model 750, it is field-upgradeable to an ES Model S800, which can hold over 55TB of storage. This system was launched in conjunction with IBM’s new midrange mainframe, the eServer z890, and has mainframe needs in mind. It has over twenty autonomic functions which allow mainframe jobs to execute in parallel. It also moves queue management to the storage away from the mainframe. The system also has several performance enablers that allow it to handle higher numbers of I/O in parallel, which should improve mainframe performance with this storage system.
It’s not surprising to see that even a disk system is influenced by the mainframe if the vendor is IBM. Mainframes have become more than a market, but have become the font of many of the best technology capabilities IBM has been able to create. We believe new disk systems are becoming less strategic to the open systems world as value is migrating to management and virtualization technologies that can handle heterogeneous environments. While management of expanding environments is a greater problem than the throughput of any one group of data, in the mainframe environment, where throughput is almost at Harry Potter levels anyway, where a disk system must be of similar breeding and provenance to be able to play credibly. IBM has focused its attention for the new eServer zSeries 890 and ESS Model 750 on a group of customers with older systems, and for customers who like the architecture, want to move new applications to it, but don’t need that many MIPS to get the work done. For this group, the one-two punch of the technology they expect packaged in a price they might not have expected, is indeed good news. The ESS Model 750 should sit well in its intended space.
On the other side, IBM is not really discussing the other implications of the ESS Model 750. This model moves high-end storage into the midrange space imitating a move the server business has been keen on for years. Once upon a time there were servers, and there were lesser systems slapped together with SCSI instead of EISA, an Ethernet card for network connectivity, and a bit of RAID 0/1 and ECC memory for reliability, and magically it was a server. Eventually this distinction became somewhat silly and server vendors started pushing high-end feature/functionality into the midrange. It appears that IBM has taken a page from their server handbook and applied it to storage. To our way of thinking, this is good news as it implies that customers will have greater continuity between midrange and high-end storage. Right now all the major server vendors offer a combination of their own storage at one level and a completely different OEMed product at another level. The pure-play storage vendors have had their hands full as well. EMC has done much to make Clariion and Symmetrix compatible and upgradeable, and HDS has done similar things except that they usually aren’t both sold through the same channels. Storage is marked by the sheer breadth and depth of vintage residing throughout datacenters and server farms. This trend to rationalize product lines and provide high-end capabilities at midrange prices will only serve to benefit IT buyers and simplify their storage headaches.