Those that follow this blog will know that I split it into two distinct pieces when I set up Words You Don’t Know (.com) about 4 months ago. Well I’m doing it again, by splitting off all the technology writing that I do (75% of what I write, to tell the truth) to a new web site, entitled The Virtual Circle (.com). If you visit this site, you’ll notice that it is not quite complete but very close to it.

The Virtual Circle

The Virtual Circle is intended to become an IT Site with some social networking capability and, it is also intended to become a platform for others that wish to post technical articles. At the moment the social networking capabilities are invisible. It will also accept other IT Analysts and other IT professionals as contributors. Again the capabilities for that are not visible either.

The point of this posting is simply to warn you that all my further blog postings relating to IT will now happen on The Virtual Circle…

Sometimes companies surf on the momentum of the market. All companies would love to be able to do that. IBM did it in the heady years of mainframe computing and Microsoft did it when Windows started on its extraordinary growth trajectory. It’s an easy situation to manage. In order to continue to be successful you just need to continue to do what you have been doing. This is where Apple is right now.

The Dynamics of Apple’s Success.

The drivers of Apple’s success are clearly multidimensional, iTunes, the Apple Stores, cool design, product quality and reliability, innovation, brilliant PR and advertising. But the foundations of it all are the products. Apple has two successful product lines, the iPhone/iPod and the Macs. These are not their only product families, but if you discount peripheral products, the only other product family is the Apple TV and it has not yet matured to the point where it is going to influence Apple’s success in any significant way.

In October, according to Gartner, US Mac shipments grew 11 percent year over year allowing Apple to claim  9.4 percent of the U.S. PC market (by units) but only about 5.27% of the Global PC market by units. There are some corporate PC sales in there, but the truth is that the consumer market is where the Mac is making its big impact. Success is partly driven by the retail stores, which account for about 20% of Apple’s revenues and are magnets for new buyers of Macs. A full 50% of Mac buyers at the retail stores are switching from PCs.

If you look at the iPhone market you see Apple selling about 20 million iPhones per year (about 7 million in the most recent quarter), with smart phone market share estimated to be about 30% in the US, but about 17% world-wide (according to Gartner).

In both markets Apple is outgrowing the market and outgrowing the competition. In terms of revenue, all Apple needs to do to succeed is to walk a straight line. For the Mac, that means broadening the channels to market in new geographical areas and continuing to open Apple Stores across the world. Apple plans to increase the number of Apple store by up to 20% next year. I was struck by a story recently of Apple opening up its first store in Paris. I was surprised that Apple had such a weak presence in the French market. The same is probably true of most countries aside from the US. Even in the UK, Apple has only a handful of stores – and its market share (at 5%) could be doubled. And, of course Apple needs to pursue its advertising strategy in new markets.

With the iPhone/iPod, the story is similar. Apple is constrained to some degree by the exclusive deals it struck with carriers soon after the phone launched. But it doesn’t do exclusive any more, and its existing exclusive deals will be running out soon. It’s hard to imagine the iPhone not doubling its current market share. Most likely it will go much higher – and the smart phone market is itself expanding, of course.

The Technical Position – PC

Microsoft could not have given Apple a better present than Vista. For a couple of years Microsoft was stuck with a clearly and obviously inferior OS to Apple and the consumer market was gradually but inexorably heading in Apple’s direction. Some momentum in the business market naturally followed.

Windows 7 may have staunched some of the bleeding. It’s a great deal better than Vista. But I doubt if it’s fixed the problem. Microsoft opened the door to Apple, and it’s too late to close it. Apple has “the initiative”, as they say in chess. The Mac OS X ecosystems is still ahead, it’s just that Windows 7 now has no obvious embarrassing defects. Windows could conceivably catch OS X in time, but it’s hard to imagine it overtaking OS X. All the innovation is still coming from Apple.

And Microsoft has nothing, absolutely nothing, to match the iPhone. The halo effect of the iPhone sells Macs.

The financial markets reacted positively but not deliriously to HP’s announced acquisition of 3Com. What does HP’s acquisition tell us?

Well first of all, it tells us that Mark Hurd, HP’s CEO, has a coherent and ambitious plan. HP is the biggest IT company by revenues (about $120 billion) and it’s a tough company to run. It has challenges. The printer market that served it so well now appears to be in decline. The PC market that was once so healthy is now in disarray. The only truly active part is at the low end with cheap netbooks and laptops, where the margins are thin and the Chinese are playing hardball.

When Mark Hurd took control from the somewhat unloved Carly Fiorina, these revenue shrinking challenges did not exist. Compaq had not yet properly been digested and the server market was in a much healthier state.  Hurd embarked on a series of acquisitions including Mercury Interactive ($4.5 billion) and Opsware ($1.6 billion)  to strengthen the software portfolio, EDS ($13.9 billion) establishing a large consultancy arm in one fell swoop, and now he has snapped up 3Com for $2.7 billion.

Knocking On Cisco’s Door

There’s no real mystery to the 3Com acquisition. HP is up-front in stating that it will compete head-to-head with Cisco. And that’s no surprise really, because Cisco announced its intention to dive into the server market about a year ago. You could say that Cisco and all the server vendors were on a collision course, but HP was probably the company in Cisco’s cross hairs, with its dominant share of the Intel server market. HP already had some networking jewels, with its successful Procurve switches and the acquisition of 3Com gives it a much larger customer footprint (especially in China) and a great deal more technology. So it’s game on.

Taking a broader perspective, what we are witnessing is a convergence in the IT industry. When we talk of Cisco, we’re really talking about the Cisco – EMC – VMware alliance, which gives Cisco a credible position as a “one-stop data center vendor.” We’ve never had such a vendor, HP is clearly heading in the same direction and it has a big consultancy arm that it can flex. The question I guess, is whether IBM, Dell and Oracle (with Sun) feels as though they have to join in the fun. We’ll be able to tell from customer behavior I guess.

Back to the Mainframe

From the perspective of the data center, we may be returning to “the mainframe days”, when you didn’t buy new computers so much as incrementally add to the configuration you had, boosting its power with more processing punch, memory or storage.

The server market is rapidly becoming a blade market which means you’re locked in to the blade cage. The networking is now the same, a cage into which you can slot switches or servers as you please. It doesn’t mean that data centers will fix on a single provider. Everyone understands that a vendor needs competition to stay honest. But it’s likely that the days of the commodity server are coming to an end.

Along with a few other people I noticed that in the wake fo the Great Recession, the server market collapsed much more steeply than anything else. The simpel truth is that virtual machines have freed up servers to deliver mor eof their power than they ever did before and buying server after server no longer makes sense commercially or organizationally

The game is changing.

An interesting debate is brewing about SQL databases and NoSQL database. So what are these things?

A SQL database is a relational database. SQL, the set based query processing language, pretty much defines what a relational database is: a collection of data held as 2 dimensional tables. It’s not that relational databases don’t allow other structures for data – the object relational databases, for example, are far more versatile in the structures you can store. But with such databases you have to extend SQL and when you do, it ceases to be a simple set-based language for getting at data.

And that’s where the NoSQL database enters in. The NoSQL term was coined by a Rackspace employee, Eric Evans. The name and concept generated a bit of a buzz.

NoSQL term is supposed to stand for “Not Only SQL.” However, in practice, it means not SQL-oriented, because SQL doesn’t define what the database is, so whether it has a SQL interface is irrelevant. Examples of such databases include: Apache Jackrabbit, Amazon’s Dynamo, Cassandra, CouchDB, Chordless, Google BigTable, HBase, MongoDB, Neo4j, Redis, Riak, Scalaris, Tokyo Cabinet, hamsterdb and VertexDB.

Why NoSQL is a Crock

The term NoSQL is really useful because it sets off an important debate on the limitations of Relational Database (see 6 Reasons Why Relational Database Will Be Superseded) but it aint never gonna be a reasonable classification of database because all it says is “none of the above”. Some of the “NoSQL databases” are for accessing large data heaps, some are for cheap indexing (that does force you to do an oh-so-expensive-in-processing-power SQL join), some are for storing documents and so on.

Try defining a ball point pen by what it isn’t. Well it’s not a fish and you can’t watch video on it and it has no nutritious value and its useless for accessing the Internet and it’s not made of uranium and it doesn’t make the sun shine. So what is it? It’s an airplane propeller, isn’t it?

The list of databases I’ve provided have been classified on Wikipedia as NoSQL. But they could be extended to include all the Codacyl databases and the in-memory databases and the XML databases and the Object-Relational databases and so on.

But there’s an important point here.

Accessing Data Is Nine Points of the Database

What makes a database any different from a file system?

Not much. A database is a coherent heap of persistent data. A file system is the same but lacks coherence, so you have multiple interfaces for accessing the files in the file system.  The persistence is important, of course, because that means that the data can be reused – so databases also tend to have capabilities like logging, replication, etc. built in, to guarantee data persistence in the event of failure. But the key point is data access.

SQL is a comprehensive access language only for data held in tables. The “relational theory”, that SQL emerged from,  is really set theory applied to sets of data, with the odd nuance thrown in. It’s viable for data held in tables, but semantically, it’s flat out inadequate.

Let me illustrate this with a simple very common example. The relational model would say that there is a one to many relationship between an ORDER-HEADER and an ORDER-LINE, there being one or more ORDER-LINEs for each ORDER-HEADER. It would also say that there is a one-to-many relationship between PRODUCT and ORDER-LINE, there being multiple ORDER-LINES for every PRODUCT. But nothing in relational theory semantically states that an ORDER-LINE is part of an order and is not part of a product. There is no semantics of composition in relational theory. None at all.

You can usually get around that inconvenience, but in some situations, such as the fairly common situation of a parts explosion, the way round generates work for everyone, the programmer, the database designer and the user.

The relational theory of data is either incomplete or horribly wrong. The NoSQL debate may help by drawing attention to that, but I suspect it won’t. In the NoSQL debate there is lots of criticism of relational database for not scaling – which in reality is criticism of relational products. There are relational database implementations that do scale well because of how they physically manage the data and access to that data.

There is very little criticism of the semantics of SQL, but that’s where the real problem is. If the semantics of SQL were any good then every new product that came to market would hold its data in a relational database. But that just doesn’t happen. Just about every new application that we see builds it own file system  – and that’s because the semantics of data access in SQL are hopelessly inadequate for most data.

The fault is in the semantics of composition. It simply isn’t there in SQL.

I just picked up from Toney Jennings blog at CoreTrace that Microsoft is including AppLocker in Windows 7.

“What is AppLocker?” I hear you ask.

It’s whitelisting. It’s a fairly simple flexible mechanism that allows the administrator to bolt down the desktop and prevent new stuff that isn’t authorized from running:

• It will stop normal malware dead in its tracks.
• Provide a simple way of preventing user anarchy where users load just about anything onto their desktops.
• It will help control semi-authorized applications like Skype, which can sometimes consume network bandwidth in an undesirable way.
• Keep unlicensed software from running (which is great for the small business and the corporation).
• Deliver a level of desktop configuration management possibly providing a means of running approved applications and implementing software updates when needed.
• Help meet security compliance standards.

This on its own is reason enough for all the corporations across the world to think in terms of acelerating their migration to Windows 7. It will be a huge boon to corproate computing and, at last, companies will be able to start throwing AV technology away and moving to whitelisting as the primary means of ensuring endpoint security. It has taken the longest time, but the AV shell game is coming to an end and whitelisting is taking its rightful place as the foundation of endpoint technology.

What Does This Mean for the Whitelisting Vendors?

The whitelisting vendors made their living by doing what Windows is now doing. The primary ones; Bit9 and CoreTrace, add value to simply locking down the desktop, so there’s no need for either vendor to feel threatened. Also security is not a Windows only thing. The big headache may belong to Windows in respect of malware, but vulnerabilities are everywhere. No versions of Unix or Linux I’m aware of have anything like AppLocker. I believe that the only OS that does is the iPod/iPhone version of OS X. So centrally managed corporate whitelisting has a place as long as there are multpiple operating systems.

What doesn’t have a place is AV technology. Now it has no real function at all. It’s time for the AV vendors to become resellers of whitelisting products.