Storagebod Rotating Header Image

Cloud

Sound of One Node Clustered

I've got a quick and dirty implementation to do at work but like all quick and dirty pieces of work; you know that the decisions you make in haste will come back and haunt you if you don't do them right. Luckily, we have some idea what the end goal should be and are trying to put together the path which will get us to that end goal without causing us too much trouble but the user is impatient to have something, so we've got to give them something.

In this case, this means us potentially building a single node cluster; we could do what the user requires now without a cluster but it will be harder in a couple of months to retrofit the final solution in as opposed to simply expanding the cluster. Building a single node cluster in this case is not hard and really only requires us to ask for an extra set of IP addresses and installing the cluster software.
And this led me to scratching my head again!

Why on earth did NetApp create a non-cluster mode for OnTap 8; this is just storing pain for users at some point going forward? A well written cluster mode has so many advantages, it seems like madness not to encourage users to go for it and it's actually a very clever potential lock-in for NetApp.

Not only that, currently, it is pretty much as painful to go from OnTap 8 7-Mode to OnTap 8 Cluster-Mode as it is to refresh your NAS device with another vendor's kit. Personally, the more I think about it; the more I think that NetApp have missed a trick on this. I am absolutely convinced that they actually over-reached themselves on OnTap 8 and promised the earth; realistically they've under-delivered. 

Most users probably won't notice this under-delivery but NetApp could have made our lives a lot easier. A single install mode which was automatically ready to be clustered was what was wanted; seamless addition of heads and seamless migration between heads. This was an opportunity to not simply delight customers by meeting their needs today but also their needs in eighteen months time when they need to add more capacity. How much simpler would it be to simply add in another head with some additional capacity and have it available for the whole namespace without have to create another one?  For all NetApp's vaunted embracing of the cloud; this seems to be a missed opportunity. 

Starting small and growing incrementally does not really seem the order of the day for NetApp; hoping that they can do better in the next releases. But if they don't, there are an ever growing number of clustered NAS devices waiting in the wings which although on the face of it offer very little over the traditional NAS devices from NetApp (and EMC) but long-term if I was a service provider looking for a long-term strategy which does take account of growth and technical refresh, I would be tempted to be having a long hard look at.

Of course, one can't really call OnTap 8, quick and dirty… 

Agile IT leads to Agility?

Leadership, management and agility are becoming important watchwords in the role of IT delivery and it is these three concepts which need to drive any IT organisation forward over the next decade or so. It is the third of these that is probably the most important.

Leadership and management can almost be interchangeable when an organisation is in a steady-state and lets be honest, most IT organisations have been in some kind of semi-steady-state for the last decade. IT Delivery became very procedural in a glacial sort of way.

Radical change in many IT organisations is simply shuffling the cards around without adding anything else in; perhaps moving what already is being done to a third party. In fact, arguably the trend to outsource can be pretty much attributed to the steady-state of the organisation. IT became something to be managed as opposed to something to lead. IT leaders who want to change things rapidly and quickly often find themselves at conflict with their own internal organisations; it becomes easier to accept the status-quo as opposed to embracing the change. 

Arguably, IT's success in becoming ubiquitous in many organisations has become it's biggest blocker to change; IT has become so large and so important that it's ability to respond to the business has become seriously compromised by this importance. Look at the sheer amount of work it takes to roll out a new version of Microsoft Office to an organisation? Okay, it is a moot point whether upgrading Microsoft Office buys very much to an organisation but lets look at the number of corporates who are still running with huge amount of risk because changing from IE6 to a later version is allegedly so complex. 

This apparent complexity has lead to an inward focus on the delivery of IT and paradoxically this focus on the delivery of IT has significantly damaged it's delivery. It has become more important to the manage the delivery than to deliver. 

However, we have already seen changes in the way that applications are being delivered and developed; agile development techniques are probably used for more than 50% of our applications internally. I suspect that this is an industry trend and your mileage will certainly vary, yet I remember when I first ran an agile development team nearly ten years ago that this was considered to be extremely unusual and that the 'Agile Manifesto' was some kind of hippie movement. 

The move to agile development techniques has been remarkably quick compared to say that of take up of object orientated techniques which can only be attributed to that there is something right and easily implementable about the techniques.  You can find the Agile Manifesto here.

Agile development techniques have some key points that I believe point to the direction that all IT organisations need to take.

  • Customer involvement in writing the stories which are to be delivered are key; every successful agile development has customer involvement at it's heart. Don't hide your people away from the Business; introduce them slowly if needed but stop making every communication between the IT organisation and the Business a management function.
  • Leaders can develop without been forced into some management cookie-cutter mould. A great developer gets to work with many different developers and gets to disseminate their greatness but there is no pressure on them to become managers or team-leaders. 
  • Teams should be allowed to organise themselves and define their own structures. They will often surprise you and if you allow them to break down organisational silos, you will find that their effectiveness is significantly increased. 
  • The acceptance that no plan survives contact with the enemy. This does not mean that there should be no plan but there needs to be flexibility and options. 
  • Simplicity and the art of doing the minimum to deliver is essential. For a long time I have preached the concept of 'Constructive Laziness'; do nothing which does not need doing and do what needs doing in the easiest manner. 

 How many organisations are ready to embrace this? I think this where the real difference between Leadership and Management will show, to implement such principles means taking up the mantle of Leader and dropping some of the comforting tropes of the Manager.  

Random Ruminations on IT Form and Function

I continue to think about the role and structure of the IT function and how it needs to change as the landscape evolves around us and here are some of my more random thoughts. 

It is important to stop thinking about the scale; scale needs to become completely meaningless; if you think about it, you will terrify yourself and fall into the traps of applying old paradigms. Forget about scale but always account for how to scale. Whatever you do today, you will need to do more of tomorrow.

I think we need to review how operational support, implementation and architectural teams interact. I hesitate to suggest that these all need to become a single function due to the health impact on some of my regular readers. But I think the divorcing of these functions has caused many of the problems that we keep hitting and will continue to do so. In order for management of environments at scale, the feedback loop needs to be closed; this means platform teams who are able manage both the infrastructure from end-to-end but also the infrastructure life-cycle from cradle-to-grave. At the very least, there needs to be regular and open end-to-end reviews involving all the players.

Too often I have seen extremely complex designs created and implemented by people who are not going to run them on a day-to-day basis slowly decay into a morass because the architectural decisions have not taken into account operational procedures, business growth and all those things which the operational teams have to deal with. 

How quickly people forget the support issues and operational headaches caused when they become no longer their problem. How quickly people forget the broken nights caused by overly complex designs which are a nightmare to scale and manage. How quickly people forget cursing the implementation team who have not created basic automation for monitoring and restart scripts.

I can point to some extremely complex storage implementations and layouts which have caused no end of headaches because no account has been made as to how to add more space. 

However even as we consider IT as a service; we also need to review the vertical services that we provide and how best to service our businesses. My favourite example is email and all the related services around email. Email is now so ubiquitous that we know what is required from it but when was the last time anyone did a requirements review of email? I hear lots of talk about Exchange 2010 upgrade projects but little about what the business wants from the service, I suspect that we have stopped asking. 

As we separate services into verticals; this does bring us into conflict at times with horizontal services such as back-up and archiving. Do you use a single back-up and archiving tool or do you encapsulate the backup and archive function into the vertical. Does this breed a multitude of products and does it increase complexity? It could do but if you are charging the business for the provision of the service and they are willing to pay the cost, this may not be important. However as you enter a service-oriented world, you might be competing with external providers. Keeping things as simple as possible would probably make sense but at times a vertical stack might make more sense than a horizontal slice.

We need to consider how we measure our success as an IT service provider and provide business relevant metrics designed to demonstrate value and not to make the IT department look good. For example, I regularly see change metrics demonstrating the number of changes undertaken successfully but what does that mean to the business. Did all those changes actually need to happen? Why are there so many changes? How many of those changes actually needed to happen because of mistakes made in prior changes? What value did those changes generate as opposed to the risk that those changes introduced? 

As we define service metrics with the business; there needs to be a mature discussion about how things like growth are handled. As IT departments, we cannot expect blank cheques but we cannot continue to provide IT services as a pure all you can eat buffet; yes, there is an always open buffet but every visit may attract a cost. 

Every IT spend above a certain threshold needs to have a worked business case and every business case should be up for a spot audit to demonstrate that it actually delivered the business benefit stated; this should not be used to beat people up but should be used to build an understanding as to what actually works and getting better at delivering business benefits.  

Yours,

A Random Bod

The Joy of Cloud

Although I have may suggested that 'Cloud is the new ITIL' in my previous blog; there is an alternative view, 'Cloud is the new sex'!

People talk a lot about Cloud, but few people are doing it!

There are many Cloud Experts, most of them have never done it!

You probably wouldn't want to sleep with most Cloud Experts! 

The best people at Cloud are too busy doing it to talk about it!

There are mixed opinions as to whether Cloud is better in private or public!

Some people would simply prefer a nice cup of tea!

Many people worry if they are doing Cloud the right way!

Choose your partner carefully before doing Cloud!

Cloud can be expensive if done with the wrong people!

Cloud can be career-limiting if done with the wrong people!

There are many types of Cloud, some are just wrong and some are just personal preference!

Your attitude to Cloud may be influenced by your early experience of IT!

Older people may be better at Cloud than you expect!

Cloud was invented in the sixties! You only think you're the first person to do it!

Your first experience of Cloud may be underwhelming!

Cloud can be messy!

How you do Cloud is probably more important than the tools!

Take your time when doing Cloud! 

Cloud can be complicated but it doesn't have to be!

Cloud with multiple partners can take careful scheduling and planning!

Only exhibitionists do Cloud in front of Windows! This does not make it wrong but it might amuse some observers.

Without protection, Cloud can lead to unintended consequences and virii.

I'm sure you can think of your own….

Cloudy With A Chance of Miracles…

Cloud will fail for the reasons just about every other IT initiative does and that is almost complete intractability of every Corporate IT department in the world and their ability to resist change. 

There will always be some reason why it can't work here. 

Of course there is lots of enthusiasm at the moment and lots of CIOs telling vendors and their peers that they are going great guns with Cloud; of course, they are surrounded by people telling them what they want to hear. I am sure that a lot of money is being spent on Cloud-enabled technology; the branding of VMware as some kind of magic cloud bean enables IT departments to say 'Yes, we're doing Cloud; look we've got lots of VMware deployed!'

These are the very same IT departments who five years ago who would have told you that they had embraced ITIL (and we can argue whether ITIL is a good or bad thing) but there sure were a lot of people claiming that they were implementing ITIL. Of course when you started to ask questions, they had implemented their own tweaked version of ITIL because ITIL wasn't really appropriate to their organisation. And then if you dug a bit more, what you actually found is that nothing had really changed and still the IT department merrily carried on in its own sweet way. 

And so I feel that most Cloud initiatives will be the same. Too many people with vested interests who are unwilling to expose their organisations to the level of change and scrutiny that is going to be needed to implement Cloud. 

I'm sure Chuck and other vendor 'leaders' talk to some very enthusiastic CIOs who are all about embracing the Cloud but in five years time will we be scratching our heads and asking what's changed? 

Cloud will be the new ITIL; just wait for Cloud v2 and Cloud v3.

But it really doesn't have to be this way however it needs some significant changes which are hard for many people to digest and get their head around. It's not all about IT people changing either.

Firstly, people need to understand the value of IT and not just the cost of IT. Let's take email for example; when email goes down, many companies seem to come to a grinding halt; so the email service has a value. The question needs to be asked what that value is, what can my company do now which it could not do without email. Perhaps it enables the company to communicate with its customers in a more timely manner and that has value. Do we actually know what that value is? Businesses need to ascribe value to the services that the IT department provide and perhaps they might come to the conclusion that they don't value some services.

Secondly, IT people need to stop operating under an atmosphere of fear; if the services provided are ascribed business value, some of the fear should be removed. Once value is understood, enhancing value is actually possible and perhaps at that point, implementing change to enhance value becomes an imperative not something to be resisted. IT people need to be proud of the service that they provide and not slightly ashamed; this is not about geek pride but a case of being able to say that we do a great job.

Thirdly, IT leaders need to stand up and say 'We believe what we do has value and we are not just a cost but we are looking for ways to improve the value of our service!' Too many IT leaders are falling into the trap of believing that their services are too expensive and hence find themselves in a constant spiral of negative evaluation. Actually most of them have little real idea as to what the value of the services that they provide actually is. 

Perhaps when you find such an organisation, let me know….I might want to work there!

Interestingly Un-Interesting

3Par finally get acquired and by Dell; mission accomplished by their bridgehead-man, Pathfinder Farley. Like love and marriage; Farley and Dell go together like a horse an carriage. 

But all joking aside, what does this really mean.

It does not mean that Dell have an enterprise storage strategy; just having a bunch of SKUs does not make a strategy. Over the past twelve months or so tho', Dell have acquired some interesting technologies.

1) Exanet – a scalable NAS product.

2) Ocarina – content aware DeDupe technology.

3) 3Par – an Enterprise Storage company.

Some people may raise eye-brows at the last statement but I believe that 3Par were really an Enterprise Storage company; or at least they had aspirations to be so. From pretty much day one, they set their sights on HDS/EMC/IBM/HP's Enterprise market and considering how hard it is to compete in that market, they did remarkably well. To gain any kind of footprint at all was an achievement and from those customers who bought them, I heard only good things. People liked 3Par, both the technology and the company. They took a difficult road and who is to say that they were wrong? The investors have got a pretty exit out of this. 

But what have Dell got out of this? They've got some good technology and some great goodwill. Can they make anything out of this?

1) Exanet – forget the NAS, it's all about ExaFS. Massively scalable, clustered file-system. 

2) Ocarina – Intelligent DeDupe which can deal with multimedia files.

3) 3Par – Enterprise Array with pretensions to federation, predictable performance and management simplicity.

Throw in Equallogic and you've got a storage cloud; it's not a especially interesting storage cloud in that it doesn't really do anything that new but if Dell can keep it very simple, they may have product as opposed to a bunch of technologies. Does it push the boundaries of what Enterprise Storage is about? No, not at all but this is Dell we are talking about and Dell are not about interesting but commoditising. 

All quite boring but maybe in a good way.

Storagebod likes this!

I really like this blog entry by Vince Westin; it absolutely confirms a belief that I have held for sometime! Application vendors have shares in storage companies! 

Poorly written SQL statements have probably driven more high-end array purchases than any salesman working for a vendor. Actually I suspect that poorly written SQL statements have driven more Enterprise IT purchases than anything. 

It is actually nice to see a vendor rather than recommending an expensive upgrade actually doing the sensible thing and looking for the underlying problems. 

Still perhaps Oracle purchased Sun because they absolutely realise that this is the case; they can now sell you both a nut but also a very expensive sledge-hammer to crack the nut. 

I suppose if I have one big concern about Cloud is that if we allow people to treat resource as abundant and cheap enough to waste; will code-bloat and poorly written code proliferate?

 I am convinced one of the major drivers of storage growth is that desktop storage is now so cheap; it can be wasted but the same is not yet really true of Enterprise Storage; a terabyte of 'Enterprise SATA' is very much more expensive than a terabyte of 'Desktop Storage' but try telling the average developer or user that? 

With Abundant Good Cheer

As we move from a time of IT as a scarce and controlled resource to a time where IT is seen as an abundant and easily available resource; we need to consider what this means to us in Enterprise IT. Is abundance a good thing and a power for good or does it bring with it issues? 

Lets look at a couple of quotes from Clay Shirky's book 'Cognitive Surplus'

'Abundance breaks many more things than scarcity does. Society knows how to react to scarcity' 

'Abundance is different: its advent means we can start treating previously valuable things as if they were cheap enough to waste, which is to say cheap enough to experiment with.' 

Now Shirky is talking about publishing, information and how we consume/create media but some of his thinking strikes a surprising amount of resonance as to what is going on in Enterprise IT today and how it is beginning to change.

In Enterprise IT, we know how to control access to resource, how to ration resource in such a way to ensure that we always have some resource saved and preserved for a rainy-day. But as we move to a world where resource is abundant; our models have to change. 

Perhaps we need to move from beyond the 'gate-keeping' mode of operation that many of us are used to where our job was only to let people into our environment if they met our very stringent criteria to a mode of operation more akin to one of the accommodating host where everyone is welcome as long as they do not impact any of the other guests.  

Enterprise IT is probably not yet at the stage where it can be considered to be cheap enough to waste but perhaps we are on the cusp of this? Or perhaps we need to be considering what needs to be done to get to that position. The Amazon EC2s of this world certainly allow experimentation at a very low cost financially. And arguably more important, the time to implement is much quicker. 

This may drive a culture of 'try often, fail a lot, succeed rarely'; as in the world of publishing, the average level of application quality may well be driven down and it will be important that we build infrastructures which can cope with failures. Badly behaved applications will need to be managed, we will need ways of spotting a badly behaved application quickly and like the good host, gently encourage it to behave better with the eventual sanction of kicking it out. 

Virtualisation can help to ensure that applications behave within limits but applications can fail in many ways and we will need to ensure that we have the tools to tidy up afterwards. This will limit the waste but allow our users to experiment and try things; yet again, like a good host, we tidy up after our guests and recycle or re-use what we can.

The IT Infrastructure Manager as 'Mein Host'…no more the grumpy chap in the corner but the genial host ensuring that the party runs smoothly yet with a swing. 

Expotition to the Cloud

One day, Pooh comes across Christopher Robin putting on his boots and Pooh is very excited for Big Boots mean that Christopher Robin is going on an Adventure. Adventures are great and Pooh tidies himself up and tries to look 'Ready for Anything'!

Christopher Robin announces that they are going on an Expedition, this is a thrilling announcement for Pooh as he's never really been on an Expotition and they sound just thing for a 'Bear of Little Brain' to be doing. Christopher Robin tells Pooh that they are going to find an Expedition to find the North Pole and the North Pole is something that you discover and there is no explanation above that required!

Pooh rounds up all his friends and them to prepare for an Expotition to find the North Pole; the message gets rather garbled and Pooh seems to forget half of it on the way round.  All the animals gather, including some who are not invited but who turn up anyway and off the friends set to find the North Pole!

The trip is fraught with danger involving Ambushes (which are not types of Gorse Bush which fall on you but are suprising all the same); Pooh sitting on thistles and Eeyore eating them. Eventually Rabbit and Christopher Robin have to consult about what the 'North Pole' is; indeed they used to know but now they have forgotten but they decide that it must a pole stuck in the ground.

Meanwhile Roo has decided to have a wash in the stream but unfortunately he falls in and although being a brave little Roo; everyone is worried about what might happen. Fortunately for all, Pooh has found a long pole and manages to use it to rescue Roo! 'Hurrah for Pooh', I say!

And what is more, as if by some miracle; the Pole that Pooh has found has turned out to be the very North Pole that they were looking for!

So what does this have to tell us about the journey to Cloud? What can we learn from the animals and Christopher Robin?

I see Christopher Robin as the CIO or at least a senior level manager in a company; he has heard of this wondrous thing called 'Cloud Computing' and has decided that his company needs to adventure towards it.

Unfortunately Christopher Robin knows very little detail and even worse, he asks his friend Pooh to communicate the message; now despite being a really rather marvellous bear; Pooh is not the world's greatest communicator and the message about what they are trying to achieve is somewhat garbled. In fact, no-one turns up to the adventure with any real clarity about what is to be achieved and what the goal really is! 

As the adventure continues and people start to grumble a bit; Christopher Robin has to turn to Rabbit who appears to be acting in a consultative capacity. It is fair to say that Rabbit like most consultants really doesn't know what he is talking about but he agrees with the client's vague ideas anyway.

Meanwhile, Roo decides to do something by himself; which is like the user who decides they are going to attempt to do something and as is sometimes the case, this leads to deep water. Perhaps like a proof of concept application which grows out of all proportion and is running away; fortunately Pooh is on hand with something he's found to rescue the day and everyone is happy. 

Even better, what he has just found is declared to be what they were all looking for anyway. In our theoretical IT company, I would suggest that they have not found Cloud Computing but a hosting company. But close enough and everyone sets off home feeling happy that they have discovered precisely what they were looking for even if they didn't know what they were looking for anyway. 

 

More Cloud Lessons from the Bear

In my previous blog, I looked at what Winnie the Pooh had to teach wannabe Cloud vendors; but the 'Bear of Very Little Brain' has much to teach us all and not just vendors. 

Although Pooh has a wide range of diverse interests, it is probably fair to say that somewhere near the top of list is Food. 

So it is a day very much like today, that we find Pooh walking through the forest humming a little hum to himself when he comes upon a hole. A master of logical and linear reasoning; a hole means Rabbit and Rabbit is Company; and Company means Food! 

So after a rather hearty elevenses, Pooh decides that he better be off (truth be known, Rabbit had run out of Food) and it is at that point he discovers a problem. Unfortunately, the consumption of elevenses has resulted in a unplanned expansion in the girth area and when attempting to leave Rabbit's hole; he becomes stuck. After pushing and pulling by Rabbit, there is a stressful moment where Pooh blames Rabbit for not having a bit enough front door and Rabbit blames Pooh for not knowing when to stop eating.

After more pushing and pulling with all of Pooh's friends getting involved; Christopher Robin comes to the conclusion, that the only solution to the problem is time and allowing Pooh slim down over time. Eventually Pooh is freed to wander the 100 Acre Wood again.

Now as a potential user of the Cloud, there are lessons to be learnt from this:

  1. When Pooh went for a walk, he did not intend to stop at Rabbit's for elevenses; it was a very opportunistic snack. Many users of the Cloud will end up using the Cloud in a very similar manner. It seems a great idea at the time.
  2. Pooh did not check with Rabbit how much Food he had and eventually consumed it all. How many people know what the limits of the Cloud are? It may appear to be infinite but is it? Will the cupboard be bare at some point.
  3. Pooh easily got into Rabbit's home because he was less weighty at that point, when he came to leave, he had expanded some what and had not accounted the limited size of egress. It is easy for a Cloud application to generate a large amount of data and getting that data out of the Cloud may be harder than you think.
  4. For Pooh to eventually escape from Rabbit's home was a long tedious process and at times a somewhat fractious process. Migrations can also be long, tedious and fractious; they also stop you doing what you'd rather be doing which in Pooh's case was eating, in your case it might be growing your business.

Many of Pooh's problems could have been prevented by planning and management of girth. Many of the potential pitfalls in the Cloud can be prevented by planning and management of growth.