Cloudy with a chance of data loss
October 13th, 2009 by ravi

Microsoft and T-Mobile have lost T-Mobile user data:

The cellphone provider T-Mobile and Danger, a subsidiary of Microsoft and one of T-Mobile’s partners, said over the weekend that a technical glitch in their computer systems would probably result in some customers losing their personal information like contact names, phone numbers and digital photos.

via Glitch Could Erase Data for Some T-Mobile Users –

What is interesting is that there seem to have been no backups to recover from? Why not? Well, everyone from Adam Smith to Richard Stallman to Joe Stiglitz has pointed out, in their own way, the simple fact that the your interests and that of those you get/buy services from might not always coincide, and in a situation where there is a great imbalance of knowledge/expertise, your ignorance can be easily exploited. This is not a point that is particular to cloud computing, but cloud computing, when consumed as prescribed, does increase your dependency and lower your control over your data.

The solution may not be to turn Luddite and shun technology, but the recent Google/Gmail issues and all that has occurred since (Apple’s MobileMe problem, the drama surrounding, the above T-Mobile/Microsoft issue) should encourage us to think harder, especially about the trend of moving away from standards, published protocols and data formats, and especially the separation of data and presentation.

Often times, the move to cloud computing (e.g: web-based email) from older technologies designed with the above criteria (open standards, separation of data and UI) in mind, is a step backward. It should be unsurprising that even as Gmail was suffering from a prolonged outage for the web interface, the standards based and client-server IMAP interface was fully functional. For those who started using the Internet after 2000, cloud computing (under which label we might legitimately include web-based services, as per their own claims — to hesitantly quote Larry Ellison: “The interesting thing about cloud computing is that we’ve redefined cloud computing to include everything that we already do”) might be synonymous with the Internet itself (just as the term World-Wide-Web is often used interchangeably with the Internet). Online services are the only ones they may be aware of, and this serves vendors in that space quite well. They would prefer that you forget the hardware device you use, the peripherals that connect to it, the power of the operating system that enables all your activities, the idea and existence of rich, powerful desktop tools (software). Again that serves their interests. Does it always serve yours?

John Gruber’s 15 minutes of schadenfreude
September 8th, 2009 by ravi

WordPress users need an upgrade to a fairly new version (2.8.3 or newer) to fix a security hole that was recently discovered. That bit of news has been dispensed via various outlets including Twitter. John Gruber, a smart and interesting tech blogger, seems to be thoroughly enjoying this business, from what I can tell. This is not the first time he has opined on WordPress, but despite what seems to be poorly concealed glee (the man has made three comments — and counting — thus far), I think he is almost as equally wrong this time around as well.

First some background:

As per my reading of Matt Mullenweg (the primary creator of the WordPress platform) this exploit is possible if:

  • You failed to upgrade WordPress for two releases (i.e., both 2.8.3 and 2.8.4 do not include this susceptibility)
  • and you have registrations enabled for your blog (how many self-hosted blogs do that? I do not know)

Gruber’s basic argument seems to be this:

  1. If you run a self-hosted WordPress blog and ignore updates, you can get hacked.
  2. This is the equivalent of “finding your home burgled and your valuables missing”.
  3. He (Gruber) does not know if WordPress is poorly designed (“security-wise”).
  4. However, whether it is poorly designed is a question that is similar to the one that “continues to rage” (for “15 years”) about Microsoft Windows.
  5. Gruber can’t recall any widespread security attacks against Movable Type, Posterous or Tumblr.
  6. Gruber doesn’t update his MT installation but he hasn’t got hacked.
  7. Therefore “the situation with WordPress is different, and clearly more dangerous, than it is on other platforms”.

In consecutive breaths, Gruber first ponders:

I have to wonder when WordPress users will start switching to some other platform.

and then disclaims:

Nor am I attempting to persuade anyone to switch from WordPress to Movable Type.

Let us take him at his word, the second one that is (i.e., he is not interested in persuading anyone to switch), and look at the argument he offers. He starts out with a truism in computer security: all software has bugs, and if you run some software that is reachable via the Interwebs, you can get hacked, especially if you don’t update the software with fixes for such bugs. If you didn’t realise that and decided to run a WordPress blog anyway, then like Gruber I would distance myself from “blaming the victim”, but I would still wonder what else you may have expected.

What you should expect, if I am following Gruber’s argument, is that if one platform (MovableType) has not suffered any “widespread” attacks in one individual’s (Gruber’s) memory, while another (WordPress) just suffered one, then that information recalled from memory “clearly” makes the second platform (WordPress) more “different” and “dangerous”. Even if we do not know if there is anything about the second platform that makes it poorly-designed security-wise.  Or in his other words, this one reported vulnerability in a version of a software two releases old is the equivalent of 15 years of [raging debate on] Windows security (or lack of it).

I am afraid that what is unfortunately made obvious by this line of argument is only that John Gruber runs an outdated version of Movable Type (that he does not care to upgrade) because he thinks, though he does not necessarily know, that it is a safe[r] platform. Which in turn leaves one hoping that he will not be in the unenviable position of blaming the victim, himself, some time in the near future. As I wrote at the top, Gruber is smarter than that… or will be once the 15 minutes of schadenfreude wears off.

Some smaller clarifications: As long as you or your hosting provider perform[s] backups, this exploit (if it occurs) does not mean that you have lost your blog content. Without doubt, restoring that content can be a painful affair, but it’s a lot easier and reliable than contacting the police to recover your belongings.

Oh and Movable Type? Here’s just what one random Googling produced (author: distler):

Remember insecure formmail scripts? How very 1990s, eh?

As if comment spam were not bad enough, MovableType includes, in its default installation, a CGI script called mt-send-entry.cgi which — you guessed it! — can be used to send email anonymously to anyone in the world.

And, no, this is not a merely theoretical issue; it’s being actively exploited by spammers.

Want more?


All these demonstrate is (not that these are the same vulnerabilities discovered in WordPress but) that software is by nature insecure. And all that rot that we all know quite well.

Virtualisation and the Verger
July 6th, 2009 by ravi

Focus your interest, if you can, on the IT manager. This once docile creature, who used to oversee the functioning of a server room or a machine room, has recently morphed, at least in jargon, to master and owner of (drumroll please), “The Enterprise Data Center“. From a 7-11 to a WalMart… lots more aisles, same old crap. And so he sat astride this beast, enjoying a newly minted cabinet-level position (CIO). An example of what came next in his evolutionary story, a veritable Cambrian explosion that should please the late Stephen Jay Gould, shall be described soon.

But first, I must clear something up: this new breed is no longer the ubergeek that you might have in mind, purveyor of magical solutions dredged up from the bowels of operating systems to cure any and all ailments of the voracious end user. This is a new man. And a new man needs a new language, and the English language, in case you have not been informed, is today taught in the School of Business Management — perhaps not taught, but more than that… researched — and this English is a language of rampant capitalisation and short span: TCO, ROI, KPI, and other 3 letter words that identify the IT stalwart as a man of heft and substance.

And so, armed with metrics and spreadsheets this new man went looking for a new problem, and sure enough he found one. Or rather, so as not to ruffle the Armani, McKinsey and Co obligingly found him one.

McKinsey, in case you are unaware of this line of profession, is one of a set of organisations that enables the outsourcing of thought. It is the Deepak Chopra to corporate angst. It is the homeopathy of business medicine. Say your car company (let us call it, hmm… so many choices… how about Ford) makes gas guzzling ill-designed cars that nobody wants to buy even if you offer to sell it to them for $2000 lesser than a competitor’s equivalent vehicle. In the days before McKinsey, you had no option but to examine your design and manufacturing processes, the performance of your top managers, your advertising and marketing… frankly the sort of tedious activities that are beneath a Harvard man like you. The advent of McKinsey was a moment of fairness in a lop-sided world lacking Live Aid concerts for Exective Vice Presidents. So you turned the issue over to McKinsey, and sure enough, within 3 weeks, the credentialed but unsung heroes within turned out a report identifying your problem: worker pay and retirement benefits. The brilliance of McKinsey’s solutions was the unerring serendipity of their prescriptions seen in light of your own innermost desires.

Sometimes, on very rare instances, when the cosmos is aligned just about right, reality takes a bite out of this entire scheme, as happened to a close sibling of McKinsey’s, by name Arthur Anderson / Anderson Consulting, also known as MBAs Sans Frontiers, whose name you might vaguely associate with that of a now black-holed hot gas giant, Enron. But one cannot build a model of the universe and go about life on the basis of singularities and exceptions.

And so, we return to McKinsey and our Windsor knotted spreadsheet chomping champion of computing.

“Listen”, says McKinsey to our erstwhile comrade, “you have a problem”.

Not one to be caught napping, our new leader responds: “I know that!”.

And then adds: “What is it? Is it domain squatting? Mitigating enterprise data mining synergy downtime? Not Intellectual Property infraction weaknesses. Please say it ain’t so”.

The McKinsey man is known not merely for finding a question to fit his answer, he is also endowed with the best of bedside manners, and with a warm pat on the shoulder he offers his alarmed client some specifics: “Do you recall our humanware optimisation strategy from last spring when we identified surplus functionality in IT services, transitioned users to a self-service model and achieved gains in payroll loss reduction?”.

“Do I remember”, grins our chap, “I thank you for it each time I climb into the XJ8 I bought from my bonus that year”.

“I think”, the McKinsey man’s manner turns brooding, “it is time to apply that same analysis to your hardware”.

“You mean?” stutters the leader.

6%“, bellows McKinsey Jr, shooting up from his chair, in righteous wrath. “6% is what I mean!” his voice trembles in outrage and incredulity.

The leader is still at loss, but feels the promise of next year’s bonus suffusing warmth through his limbs. “That doesn’t sound right!”, he responds in indignation.

“It’s sacrilege”, says the manager-whisperer, “and it is the current utilisation percentage of your enterprise data center“.

“That’s bad”, nods the new organisation man, “…. right?”.

“Terrible”, comes the assurance, “just terrible. Here you have all this spare CPU capacity going waste.” (he considers adding: “There are children in India who don’t have a single CPU cycle and you are wasting 94% of it each minute, each hour, each day” — but then decides against it. The McKinsey boys are known in the underworld for their soft touch).

“Should we run more apps”, offers our hero, “we can run more apps. We could convert the code base to C++ or Java and that’s like 4 times the CPU cycles. I could call the boys up at Microsoft — their new product Basic Limited Office Add-on Tools gets like the highest bars in all those performance benchmarks”.

“No, no”, McKinsey Jr is not satisfied, “we cannot always fall back on Microsoft to solve our problems, and we are moving away from C++ as a revenue generation source”.

Allowing for a pregnant pause to underscore the heft of his proposal, the hired brain leans forward.

Virtualisation“, he intones, assuming the voice of the Dalai Lama, whose teachings, it had turned out, were invaluable to the aspiring technology leader, and were made available for free with the purchase of a full set of Taylor Made clubs.

“What?”, the confusion was understandable, for the word “virtue” had not enjoyed any prominence in prior consultations, “Virtual liaison?”, the manager had always liked the word “liaison” and his hopes shot up.

“No, no”, muttered the peeved boy genius. “Virtual-i-sation. What we do is we separate the OS from the hardware… introduce an abstraction layer between them.”

“Ah, and that fixes the CPU under-utilisation”.

“Well yes, but…”, the man from McKinsey was a decent sort and one did not just blurt it out like that, “what you really gain is the power of virtualisation. Think about this for a second. You take all your hundreds of servers, and you buy one big honking server — with lots of redundancy of course so the hardware doesn’t fail — and migrate all of your applications and services to virtual machines running on this big honking server. Now you have a single point of failure … no I mean you have…”, he dug deep into his mind searching for the bullet list that he recalled from slide 3 at the training session, “you have improved manageability, yes, improved manageability, consolidation, efficient resource utilisation, expandability. And, and… you can migrate your virtual machine with all its services to another server at any time. That’s the beauty of virtualisation. You can just plonk the VM on any of your servers”.

“Woah”, the manager falls back reeling in amazement.

“That’s right”, the management guru-in-the-making moves in for the kill, “let me just say that again: you can consolidate all your servers into one server, saving hardware cost, but then at the same time you can distribute your services across many many servers by virtualising them”.

“Awesome… I have always been worried by Moore’s Law. This is just, I mean, just amazing technology to solve it. I think we should bring in my sys admin, er I mean, IT architect”.

The call goes out for the IT architect and in walks a bearded bloke with a T-shirt depicting some sort of a red devilish figure holding a trident.

“We need to implement virtualisation for our data centre”, declares the pioneering captain of enterprise computing.

The McKinsey man steps in, “Imagine“, he stretches out his arms wide, “all your tens of servers consolidated into one single server”. His arms implode, his hands meet to form a cup (within which we presume the single server resides).

But…”, the IT architect starts to object.

“I know”, nods the consultant, “I know. How? Yes? Virtualisation. That’s how”.

But…”, the IT architect persists.

“I know”, smiles the consultant, “what about security? That’s your concern, yes? It’s all built secure from the bottom up. AAA certified by Moody’s Computing”.

“But”, says the IT architect, finally squeezing in a larger part of his sentence, “we don’t have tens of servers. We have three servers which implement our core services including DNS, Mail, web applications, …”.

“Three?”, the McKinsey wunderkind turns pale, “Three? That can’t be. A study by our team has demonstrated that an enterprise of your size has a computing hardware resource overhead of 43.5″.

“So, we can’t do virtualisation?”, the boss is crushed and puzzled.

“Clearly”, McKinsey’s voice took on a steely tone, “you are not yet ready for advanced technology such as this when you are not able to even meet industry standards on such a simple metric as computing hardware resource overhead”.

“Can I go now”, asks the sys admin, “we have a performance issue on one of the servers that you said we had no budget to upgrade, and I have to move the DNS service to the secondary, while I add memory scrounged from other systems to the first server”.

“Oh”, the consultant is curious, “how will you migrate your … what did you call it? … DNS service?”

“Change the config file and bring it up as the master”, replies the sys admin as he heads for the door.

“Oh my God”, the whiz stares at the IT geek as if he were some prehistoric monster, “I wonder if you had virtualisation what you could be!”.

A verger?”, quips the hacker as he waves goodbye.

»  Substance: WordPress  »  Style: Ahren Ahimsa