PGTS Humble Blog
Thread: Microsoft (Decline Of)
|If at first you do not succeed ... You must be a programmer!|
Hoisted On Their Own Petard
Chronogical Blog Entries:
Date: Thu, 30 Jul 2009 23:07:00 +1000
Already the speculation has started. Bloggers, online journalists and pundits are asking -- What will come of the recently announced joint venture between Microsoft and Yahoo? And some of these pundits are already declaring "At Last! Some Competition For Google!". Now, as your humble blogger has opined on many occasions, whatever challenges Google faces, and there will be many, they need not be overly concerned about competition from any entity using Microsoftware for Search Engine technology. This is because of the inherent performance bottlenecks that come with such software.
As usual, with such speculation, the bloggers have over-looked the technical details.
Let's consider some of the obvious disadvantages about trying to build a complex high tech application (like a search engine) with Microsoft bloatware:
Microsoft Server Farms are expensive and inefficient. And there needs to be a lot of servers. Consider, for example, corporations who have adopted Microsoftware throughout their infrastructure. Often you will see that they have a separate MS exchange server and separate Domain controller for each State (Branch). Then there might be a separate FTP server for each application. A separate LAN server for each department. Oh and don't forget the separate Virus scanners that are busily scanning email and files on various servers.
The myth propagated by the Committee For A Microsoft Flat Earth, is that the reason for so many individual servers is the low price of hardware... "It's because today, hardware is so cheap!" - Or so the theory goes! ... In fact, the reason is more likely because Microsoft machines are often only capable of performing one server task. And even then it will perform that task very, very slowly!
And if that weren't bad enough, all that extra hardware translates into a bigger budget. You will pay more for software ... Well you have to buy each license. And the additional costs don't stop there. There are more boxes, with more power supplies. More towers. Taking up even more floor space. More cables and wiring. More cooling. More sparkies. More infrastructure and personnel generally.
And the Microsoft architecture is not well suited to automation. There are some great tools to automate scripting, but these are all third party Open source products (like Activestate Perl). The basic system stuff is more labour-intensive. You must have a lot of trained monkeys, clicking little buttons which say "Yes - I am sure" ... Or "No, I'm not sure!". That's more staff! ... More MSCEs ... I think you get the picture ... It's all good for the labour force, but it's not really very efficient.
And the bad news doesn't end there. When it comes to system design, it really starts to get ugly, with many managers looking after each programmer, passing design specs down a vertically integrated chain of hierarchy. Contrast this with the Open source distributed model, where many developers and system testers co-operate using the very same systems that they are developing and the cost disadvantage of the Microsoft model will really bite you, where it hurts most.
And finally when the rubber hits the road, the performance of the end-product can be poor, especially if the design calls for significant data throughput. Microsoft machines perform great when it comes to producing eye-candy. But when it comes to shuffling data ... Lots of data! ... The performance is -- to put it mildly -- Very, very ordinary.
Of course the sluggish performance and compartmentalised architecture doesn't really matter if you are creating a payroll system for "Northwind Traders". But, if you are creating a massive climate modelling system using parallel processing ... Or an integrated transaction processing backend for a multi-national financial institute ... Something which would require clusters and massive throughput ... Then that poor performance will land you in a world of hurt.
And search engine technology is just such a high-end application. The numbers alone would make the average system architect's head spin.
That's why MSN was a flop, and Live Search was a flop and the recently re-badged release of Bing was also flop. It was a pretty, frilly and well decorated flop ... But a flop nonetheless. The performance was poor! The search queries were limited, and often out of date. But the eye-candy was great!
And the poor results do not result because Microsoft programmers are a bunch of idiots ... Microsoft can afford to hire top quality programmers ... And often they do! No, it's not the programmers ... And it's not the algorithms ...
It's the performance, Stupid!
Or more correctly, lack of performance! Especially at the back-end.
This is where it gets interesting.
Although us members of the general public don't have access to all the details of the agreement between Microsoft and Yahoo, it would seem that Microsoft are going to be providing the search technology and Yahoo will be handing their user base to Microsoft, on a platter.
This seems to be an amazingly bad deal for Yahoo! Has Bartz already commenced working for Microsoft? Or does she have a hidden agenda? Time will tell ... But your humble blogger will confidently predict that this is the beginning of the end for Yahoo.
The most intriguing thing about this deal is the division of labour ... Which is so counter-intuitive! Yahoo's open source technology is so superior to the "Bing Bungle" that it obviously would have been a better choice for the search engine. But it now seems that modern day technocrats bureaucrats who currently run both these organisations are clueless about the minutia of daily operations.
For years Microsoft have pumped their second rate software down their distribution channels and if it went belly up (like ME or Vista) -- they could walk away from it. They have slyly, but effectively relied on subversion of open standards, spin and inducements to "upgrade", and all the while, promoted the idea that they are a provider of "Quality Software". Their unfortunate customers have born the brunt of the obvious limitations of the Microsoft paradigm.
This time it is different. And unlike last century, there is no longer a hated "Big Brother" (IBM) for Microsoft to fulminate about. Now Microsoft is the "Big Brother" at the top of the pyramid and this time it is not one of their unfortunate customers trying to build an application. It is Microsoft themselves who will be building a core business system using their own product ... This time when the Microsoft troops boldly ride forth into the valley of death, they will do so carrying all the excess weight of their own development paradigm. Your blogger will humbly speculate ... If it took an entire decade to deliver a working version of Longhorn (AKA Windows 7) -- How much longer will it take to deliver a competitive, multi-platform, parallel processing, search engine technology? Especially considering that Micorosoft now seem determined to eliminate unix servers from their backend processing.
By now the great Führer of Microsoft should be regretting that he didn't take note of the fate of the other Führer, last century ... His general's dreaded the opening of the Second Front that became the turning point in the last major world conflict. And in some ways the bold but foolish commitment to Bing with their own technology may become a second front for Microsoft. Already, even though Steve Ballmer will not admit to it, Open Source is causing Microsoft pain. They have been forced to sell Windows XP for $15 per copy in order to counter the threat of Linux on netbooks... And contrary to the trite nonsense peddled by many so-called "Tech journalists" ... It is Open Source, not Microsoft that just won't give in. And it is Open Source, not Microsoft ... That will "just keep on coming".
So, how do you compete with a product that is free and maintained by skilled volunteers? Well, dear reader, it's like trying to compete with oxygen by holding your breath.
This time Microsoft are actually going for broke! And your humble blogger will be watching this enterprise with keen interest and much anticipation. Because this time, Microsoft may not be able to walk away from the resulting train wreck. They will be lucky if they can be carried away on a stretcher ...
Now, Bing must crash through ... Or Crash!
Will Bing be hoisted on its own Microsoft petard? Let's just wait and see, shall we?
Update: Not long after I entered this blog, I heard the amazing news that Microsoft had doubled the cashback for Bing!. This prompted me to hastily re-write a small article and publish it here.
How can we take Microsoft Search seriously if they have to pay people to use their search engine? How long can they keep it up? ... Questions which will be will be answered in the fullness of time.