OS/2 e-Zine! - Now on CD (Click here)
Aurora, the Solution - by Bob St. John
OS/2 e-Zine!

Go to a Printer Friendly version of this page

Bob St. John is a Business Development Manager at Serenity Systems International, a Managed Systems company

Related Articles
Unhealthy obsession with NT
Making Money With OS/2
Arguments for a New Client

Lash Back! - Join in the ongoing public discussion with our interactive forum. Be frank, be vicious, you can even be anonymous.

- or -

Blast Back! Send a private message directly to Bob St. John with your thoughts:

Contact OS/2 e-Zine!

Sponsored links:
The OS/2 Supersite

Summary: Can Aurora make a difference? On the eve of its release, Bob studies Warp Server for e-business' chances of success.

Can Aurora Make a Difference?

Will people buy Aurora over NT Server? In my opinion, IBM NCSD (Network Computing Software Division, responsible for OS/2) has certainly given cause for people to consider it. The gamma product is excellent both in quality and functionality. OS/2 organizations with Windows clients, or such platform agnostic organizations as you may find, will love Aurora because of the strong feature support for Win32 clients. But, as the discussion from last month indicated, a mind set exists which compels people to use NT, even when all evidence suggests that it is a bad idea.

Yet OS/2 Warp Server for e-business (Aurora to its friends) has a chance because it's being positioned as something more than just an application/file and print server to workstations. Hopefully this will get Aurora some press and attention in an environment in which image and "buzz" are important. Not to mention the fact that some organizations may actually want that functionality. An organization which considers itself a Windows shop has already determined the NOS. IBM Suites for NT tends to demonstrate that "Windows shops" take their vendor allegiance very seriously.

With this in mind, IBM puts a new spin on an old friend by adding the "e-business" moniker. But it isn't as shallow as the name. NCSD is serious about positioning Warp Server as a web server and a Java application server. And, YES, I think this latest Warp Server release is still the superior application server, database server, and file and print server, as well as managing the network and being easy to manage itself. This is now especially true for Win32 clients, if anyone cares, and I'm sure that some do.

The challenge is ....

who is going to sell it outside of the OS/2 enterprise accounts?

You mean it's a Desert Topping AND a Floor Wax?!?

The Aurora product positioning reminds me a little of the five blind men who touch different parts of the elephant and then describe the animal. What I mean is: Aurora strikes me as an attempt to deliver a solution by selecting different components that actually solve problems and integrating them into a product. A refreshing approach, at variance from simply delivering good technology and hoping users will crack the code on how to use it to its best advantage.

Of course, there is a question as to whether this was by design or happy coincidence. Random chance, or divine plan? Since we are quick to put blame, I'm going to give some credit. But it will still be up to users to demonstrate the value. IBM is still providing the tool. In this case, it's closer to a complete set of tools packed into a tool box.

Did IBM correctly define the business problems? Are they delivering solutions that people care about, that people are asking for? Who is the customer? Is the product priced correctly for the customer and is IBM prepared to market appropriately to that target customer? In short, if aggressive marketing is required, does IBM have the will and the funding to execute a successful marketing campaign? And related to several of these: How effective will management and administration be without a client system using the same kernel technology? The NT Server / NT Workstation model seems effective. I think an Aurora Server / Aurora Workstation would be more marketable than an Aurora Server / Merlin Workstation combination for several reasons. Not the least of which is the perceived investment by IBM and the willingness to participate in the market.

But before we go there, let's pause and briefly review how the product features may be received by users.

Even before getting there, you need to install the product. All I can say on this topic is that everyone I spoke to who installed the gamma encountered no installation problems. Amazing! True, the sample was small and these are people and systems that generally were running OS/2 Warp Server before. But the feed back was universally positive. And I expect that many of the initial Aurora users will be Warp Server users putting it on machines which were running Warp Server before. Still, the installation process has always been an issue and a good one is important, even among experienced users, because it's the user's first "out of the box" experience.

Next, the top vote getter seems to be the Journaled File System. In a conversation with one ISV about a soon-to-be-released application, I commented that Warp Server Entry could probably support some smaller environments. His response was "I gotta have the e-business server. My application will be using UDB (DB/2) and I want the users to have JFS ...it's worth the money, even in small accounts."

Part of what struck me here is that this application will probably be marketed as an integrated vertical application, hardware, NOS, application and database. So, using Aurora over Entry could be bumping his SRP by almost $1,000. But in his mind, that's the value of JFS accounts using UDB, even smaller accounts.

File System Support

There are several things here that strike me favorably off the bat. The computing environment is putting more and more emphasis on networking -- all types of networking. In my mind, to effectively manage the environment you want the ability to have specialized client systems, from fat to thin, and managed by the server to the extent possible. This assists in managing the environment, deploying software, and recovering from failures.

Not surprisingly, I also favor specializing servers by task. If you gain acceptance to evaluating servers this way, Aurora has a better chance. You can define particular server roles for Linux, NT, and Aurora. If you can get that far, establishing Aurora a superior choice for boot server, domain server, file and print server, application server, even web server, stand a good choice.

With this in mind, I favor the ability to be flexible, dynamic, and to implement change easily. Aurora seems to deliver this and one of the key areas is the file system support where we find new capacities, a new file system, and the Logical Volume Manager, (LVM). To me, those are the stars. There is also the Network File System (NFS), which has value if you need it.

Journaled File System (JFS)

A technology used by the IBM AIX group for awhile, this is way overdue for OS/2 Warp, but better late than never. Using FAT or HPFS file systems put users at the mercy of CHKDSK to correct corrupt disk addressing structures. This can be time consuming on a workstation, on a server it can be intolerable.

Keep in mind that Warp Server used to support a file size of up to 2 gig and a partition of up to 64 gig. With JFS both these limitations are raised to 2 terabytes. Can you imagine running CHKDSK in such an environment? But the JFS is continually writing a journal of transactions. This means the file system can be restored by replaying the log and rewriting the records for the appropriate transactions. Where CHKDSK might have taken hours, JFS can restore the system in minutes, perhaps only seconds.

Now we go back to the ISV with the application which utilizes DB2. His application, expected to launch this June, is a Time Clock. His Time Clocks are diskless PCs, little more than flat panel monitors in appearance. It tracks people, tasks, rates, and so forth. It integrates with Payroll and, in a billable time environment, can calculate rates, tasks, and projects for billing. It integrates into a new accounting and finance package. All running on DB2.

We can now see the wisdom of being an Aurora only solution. One failure would justify the product. Imagine selling this application and telling the customer that the type of failure they just encountered is rare, but they should not anticipate losing their information. However, it may be a couple of hours until they can use the "time clock" again, so they should take copious notes about what everyone is doing! Imagine the customer satisfaction.

To me, that is a marketable business solution that puts real meaning to the phrase "improve the availability of your system". That type of phrase is meaningless motherhood marketing. But talking to customers who bill people based on time, tasks, and rates, and selling them a system which can automate all that with touch screens or a magnetic stripe reader for id cards, and telling them they can have fast failure recovery ... that is meat. That solves problems. In marketing terms, that "relieves pain".

Disk Management

Logical Volume Manager (LVM), in combination with JFS, just suits me to the ground. This is because I'm just not all that good at figuring out things in advance. I find I need the ability to support changes and additions on a pretty regular basis, so LVM features can simplify my environment considerably. Keep in mind that words like "simplify" also tend to mean "control costs". This flexibility is also required by service providers supporting servers at customer sites. Things change. Aurora just makes changing things simpler, easier, and faster. And that also should be cheaper, because time is money.

Aurora supports FAT, HPFS, HPFS386, and the new JFS and NFS, so there is a nice stable migration opportunity. In Aurora, there is LVM replacing FDISK, and providing new functions.

  • Logical drives can span multiple physical drives and there is support for sticky drive letter assignments and permanent assignments. This means that drives can be moved or added without changing drive letter designations, leaving path information consistent.
  • JFS supports the expansion of a mounted and actively accessed file system. If a file begins to run out of space, the administrator can increase the volume size without disrupting transactions. No need to stop operations, back-up, reformat the partition and restore the data. JFS and LVM support dynamic resizing of the partition. In fact, you can write a routine to do just this, as long as the space is available. Simple, fast, even "automagic".

    Let's go back to the ISV with the financial package running on DB2, handling time clock routines, payroll, billing, even inventories. DB2 sends an "out-of-space" error message. A routine is written to intercept the error message and call the LVM API to increase the partition. A page is sent to the administrator so the change can be documented and additional modifications and changes can be considered.

The enhancements to the file system, and associated benefits, are the most important changes for me. There are some other base changes which are nice.

  • Support for physical memory goes up from 2GB to 4 GB
  • Support for virtual memory goes from 512 MB tyo 3 GB
  • The maximum processor support, 64 processors, is now supported "out of the box" and not as a separate feature.

Overall, Aurora picks up some nice features and device support which make it better suited to take a place in the current computing environment. A reliable system with improvements which make it easier to administer in a dynamic environment.

Support for Windows

Aurora's support for Win32 clients is extensive, not all of it is "new news" with Aurora. What makes the story so much better now is the support for NT Server and so many diddly administrative tasks which had to be done twice when Win32 users were using Windows and OS/2 servers. This type of administration is much cleaner with features in Aurora. Of course, Aurora also has excellent support for other clients, DOS, Win3.1, *IX. and, oh yes .. OS/2, Java, Network Stations, and so forth.

If a network has the proper organization infrastructure, these features go a long way to allowing a user to roam about with one secure logon and, from anywhere in the network, access any authorized resource in the network without creating a mountain of adminstrative management.

In my opinion, you could make it simple by not using Windows. But the mixed OS/2 and NT clients and servers can be a technically elegant network with the features of Aurora. This will allow an organization to make server decisions based on the job to do be done, not on what type of user works in which part of the network.

Aurora also includes IBM Neighborhood Browser Enabler 1.0 which allows clients with Microsoft's Network Neighborhood to access resources on Aurora. (I'd love to hear from users in mixed or even Windows environments who can comment on the value of such features and if they effectively influence "buy" decisions).

And then there's JAVA and "the WEB"

This could be Aurora's opportunity. This could be the differentiator and the opportunity to get some attention. The issue may depend on the strength of demand for "e-business" solutions, how well the role is defined and how well Aurora fits into that role. I may be just showing my ignorance, but e-business to me is still largely hype and buzz at a time when Aurora needs firm footing and traction.

This is not to say that there are not significant "e-business" opportunities. But e-business seems to be defined here as some type of web integration, browser based transactions, and "applets" and "servlets". Some folks are making serious money here. Why not Aurora? Or why Aurora when AS/400 and AIX fit this role very well, too? This is a serious issue because much of the e-business space inside IBM involves Lotus and Tivoli, two divisions which had pretty much given up on OS/2. Is IBM really ready to promote it and and support it, now?

If "Yes!", then Aurora has a real chance because this is a space where Aurora can blow the doors off NT. Microsoft doesn't want to compete in this space. It would rather create a parallel universe capable of engulfing it. That isn't likely to happen. I'm not convinced IBM is ready. But I have no doubt that NCSD is ready. Question would be, is NCSD enough?

When Serenity Systems started to design our Managed Server, we decided on Warp Server SMP because of its reliability. We also included Domino Go and Apache, because we wanted people to be able to quickly and easily build web pages for any size account. The intranet model is a great one for small accounts and web pages are an easy way to create, store, and retrieve information. Web pages and browsers are becoming very simple interfaces which are familiar to every business user. So including these features in Aurora makes sense.

At Serenity Systems, we felt that a desktop needs complete messaging support and collaboration, so we added RelishNet, FaxWorks, PMMail/2, Communicator, and Hethmon Brothers Mail Server . Soon we plan to do some telephone and voice mail integration. In other words, not a technology, or an application, but a solution.

My point isn't so much to promote Serenity Systems as it is to explain that I think in terms of a business solution. IBM tends to deliver products and technologies and hope that users will crack the code and turn it into a solution. Lotus, or more specifically, Iris, with Notes, came a lot closer to delivering a solution. Aurora could be a great Notes environment if Lotus ever becomes interested in OS/2 again. Should Lotus commit to an R5 client for OS/2, I'll attribute more value to Domino Go and other Lotus components. I think Aurora and OS/2 would derive a great benefit from a Notes integration and give Microsoft cause for concern. But it isn't likely to happen.

The integration of Domino Go and WebSphere with Aurora represent a direction I'll call "inclusionary". This is a reason to feel good about Aurora. What I mean is, NCSD is including things in Aurora that make sense. The exposure is that such "inclusion" can hurt ISVs, but that really can't be a concern now. Having a successful OS/2 product is the issue. ISVs will find a way to make money off the improved market, even if all that improves is the receptive attitude.

In line with this is WebSphere. It's a little like being able to put RISC drives in Aurora so *IX users can access the resources over the network. It replaces CGI scripts with improved performance, security and scalability. Plus it employs open standards such as JDBC, CORBA and Java. If you are serving Java applets to thin clients then this could be an important feature. But even so, it would seem to be a plus for people who are already favorably inclined to use Aurora, not to drive a "buy" decision.

I think Austin (IBM NCSD) deserves congratulations for the marvelous job they have done making OS/2 a premiere Java platform. In line with that, Aurora comes with the best performing JVM in the industry. I fully expect that NCSD will enhance this side of Aurora, including Communicator, because they cannot afford to let Aurora fall behind on the Internet arena where the web model and Java can be the difference.

And this is where I would look to see NCSD continue to invest in Aurora, enhancing web related products and technologies. If NCSD wants to put some umph behind marketing activities, the best thing they could do is talk up planned enhancements. This is not a time to be cautious or secretive. A lack of discussion will be seen as a lack of interest on intent on behalf of IBM.

Systems Management

Just what is "Tivoli Ready", anyway? Has Tivoli fallen on hard times and decided Warp Server may represent an opportunity? My impression of Tivoli is close to my impression of Lotus. A division of IBM who was inclined to say to NCSD, "If you think it's important, here's the code. You do it."

Remember Netfinity Server, once the property of the PC Company, then turned over Tivoli when IBM acquired Tivoli, then became TME 10 Netfinity Server? After awhile, Tivoli gave it to PSP who kept it alive because Warp Server users needed the functionality. As I read through the marketing materials, what is back in Aurora? Tivoli and Netfinity Server 5.2, with comments about tracking hardware and software in the network.

Years ago I remember Trantor Technologies putting a slick demo on their site. It used Netfinity, Net.Data, and DB/2 to dynamically build web pages which showed inventories of hardware and software in the network. A true business solution, applying product and technology to support business. I thought I was seeing the rebirth of such righteous capabilities. But now, I'm not so sure. About all I can really find in the documentation is that an Aurora server can be tracked, and a page can be sent to the administrator if a failure occurs. Sort of a scaled down Netview.

Tivoli Management Agent (TMA) 3.6 helps customers remotely distribute and manage software. But so does WiseManager (shameless plug deemed appropriate) for about one-tenth the cost. I admit a bias against Tivoli. I expect that someone will pick me up and put me down by correcting my impression. That's fine. I look forward to the education. It may help my ability to sell.

A potentially important feature, depending on how many people open the box, would be the Lightweight Directory Access Protocol (LDAP) Client Toolkit for C and Java Version 1.0 (Notice a lot of 1.0s? Is that a good thing, or a bad thing?)

  • This allows applications to access, search and update LDAP servers using LDAP V2 or 3 protocols
  • Includes sample client programs and libraries
  • Supports directory servers such as X.500 and DCE (Distributed Computing Environment)

Personally Save and Sound, PSnS, Backup and Restore, V6, is included to provide automatic file backup to a wider range of devices, including removable media, with fast recovery of deleted files. PSnS has always been something I thought rounded out Warp Server very well. Though I would agree that many environments should consider the stronger capabilities of CDS's BackAgain/2.


Aurora is a no-brainer upgrade for most Warp Server organizations. This is because many of these organizations are large accounts which have "upgrade protection" programs with IBM and will pay nothing to get it. Or, more precisely, these customers already paid for the upgrade. It's not like no one knew this was coming. After all, in 1997 Donn Atkins promised this server (and the client) for delivery in 1Q98.

For the rest of us, we're going to find Aurora to be a world class upgrade to Warp Server, carrying a world class price tag of $1,700. Upgrades are $1,300 if you've already got Warp Server and Warp Server Advanced. Competitive upgrades are $1,400 and User Access licenses, when required, are no walk in the park either.

Even so, Aurora provides solid enhancements. It's a quality product with a great new file system and Logical Volume Manager features and better overall industry support. It is worth the money. OS/2 Warp Server users will likely upgrade. Selling it outside, in the real world, will be difficult for reasons which have been discussed here on a regular basis.

Frankly, to sell this server to most business accounts would require a client announcement. Not because users need the client but because resellers may not buy into this product unless IBM demonstrates a serious commitment to the product. As we have often heard, if IBM isn't taking this seriously, why should I?

* * *

I'm going out on a limb and say that Aurora is going to get some qualified success. It is going to get some good press because it deserves it. But it is only going to those people who already want it, because for the time being I don't see anyone who feels it is their job to sell it. Some effective, laser guided, pin point, marketing could make a difference. Now, I would like to hear your views. You can discuss them with me and other readers in our interactive forum.

Copyright © 1999 - Falcon Networking ISSN 1203-5696
May 1, 1999