My recent post about the origins of Symantec’s Enterprise Vault product provoked a flurry of email conversations with many people, including the redoubtable Nigel Dutt, ex-CTO of KVS, the creators of Enterprise Vault.
Nigel and I have butted heads over the year, mostly because he believes that English rugby is a superior life force. He is wrong on that point, of course, but he’s been right on a fair number of other points, many related to technology, over the years. One of his best decisions was to convince a collection of ex-ALL-IN-1 engineers based in DEC Park in Reading (truly a building that offended the eye no matter what the viewing angle) who had been made redundant by Compaq to join him on the journey to take the Digital Enterprise Vault product and launch it as the first archiving product for Exchange.
Many of the Microsoft folks who sell the benefits of Exchange 2010 archiving are just plain rude about “stubbing”, the practice of leaving stub items in user mailboxes behind after they are removed to a separate archive. Microsoft says that it’s so much better to move complete items into an archive mailbox that can be managed as part of an overall data storage strategy, including the ability to satisfy compliance requirements. This is a fair argument and certainly Microsoft has achieved a nicely integrated feeling across all of their compliance features in Exchange 2010. However, it completely ignores the salient fact that “stubs” were really the only practical approach that third party software developers could take when they integrated their products with Exchange in the early 2000s.
Stubbing was an effective mechanism then because it allowed administrators to offload data from stressed mailbox databases running on expensive storage to cheaper storage running in a HSM model. Remember, the ESE database engine that underpins Exchange has really only become super-efficient through the extensive tweaking that Microsoft did in Exchange 2007 and the complete schema overall and further tuning in Exchange 2010. Looking back, Exchange 2000 was a bit of a dog when it came to storage performance and Exchange 2003 isn’t too much better, so stubbing was a very good approach to archiving then. It’s less attractive now but that’s the way of the world – everything improves and changes over time.
Speaking of ALL-IN-1, I was delighted when Nigel pointed me to Skip Walters’s blog, where I found some interesting information about how the ALL-IN-1 product was created through direct customer demand for Office Automation software in the early 1980s. It all seems so long ago now and the thought of creating a $500,000 system running on a VAX11/780 computer that could support up to 24 time-sharing users (8MB system) using a mixture of email and word processing seems really antiquated, which of course it is. The genius of those who created ALL-IN-1 was to see how they could meet customer needs in a very elegant and easy-to-use fashion.
Microsoft took ALL-IN-1 seriously enough to regard it as both a direct competitor and as a target for customer migration when they launched Exchange 4.0 in 1996. I can recall the almost surreal experience of reading the briefing documents for Microsoft management that explained market conditions, competitors, the advantages of Exchange, and how migration could be done. As I had very fond memories of ALL-IN-1 and had written two books about the technology, it was both enlightening and depressing to read the black and white assessment of its strengths and weaknesses from the perspective of Exchange.
Oh well, time marches on… We’ll probably be having similar conversations about Exchange in 20 years time when we’ll probably be communicating through mental waves facilitated by implants just behind our ears and a constant wireless connection to whatever the Internet evolves to be at that time.
Tony, I know you’ll be delighted that DEC Park was totally flattened a while back and even more so that the car park is (or was) used for overflow parking for London Irish rugby matches!
I remember doing the ALL-IN-1 futures pitch at a US DECUS in (I think) 1996 and saying maybe we would port Office Server to NT, which got a big round of applause. Not long afterwards the product manager passed on a message from our VP to the effect “don’t ever say that again”, so the message had reached HQ very quickly and obviously a boat had been rocked.
Ah yes, the famous ALL-IN-1 OfficeServer for OpenVMS port to Windows NT 3.51… that would have been quite some project. Given the importance of the Alliance for Enterprise Computing to Bob Palmer and the rest of his management team and the inevitable squawks that would have come from Redmond, it’s not altogether surprising that such a project would have been swiftly canned. Pity that…