Against this backdrop, the importance of understanding how cloud computing meets the needs of your organisation and the extent of its impact on performance and cost becomes paramount. This is compounded even further by some unique issues that have come about due to the nascent nature of cloud computing technology.

The most significant of these is vendor lock-in.  Cloud computing has essentially negated the standardisation and common protocols brought about by service-oriented architecture over the previous decade.  It is a relatively immature technology and, therefore, standards are still being formed. Customers are not yet demanding a high level of interoperability and it is obviously not in the best interests of a vendor to supply an interoperable solution that enables a customer to jump ship. This current state of play is unlikely to change anytime soon in light of the amount of capital vendors have invested into their cloud platforms.

Therefore, before you commit your hard-earned Dollars (or Pounds or Euros) to a service agreement that you might not be able to back out of, it is vital to know that such a solution will actually meet the needs of your organisation. The key here is to ensure you can identify and then accurately track the right KPIs (key performance indicators) for your business – after all, with cloud computing still so immature, there is no consensus on what factors your organisation might find important. If we take trading systems as an example, most cloud providers will be happy to provide you with stats on how much capacity you can have, at what cost. And true, capacity is important to understand – a high volume trading system needs plenty of reconfigurable capacity to deal with peaks and troughs in trade-volume demand. But arguably, the latency of such a system is an even more important KPI to accurately track. In 2012, capacity is cheap, but latency will cost you business.

In such a scenario, analytics can provide your organisation with the visibility it requires at each step of the cloud-transition journey. Starting with an accurate baselining of your current capacity and performance, which allows you to provision your cloud solution with a view to balancing cost versus requirements. Then informing the scope of a cloud services contract, assessing the impact of change that the migration will have on your systems and processes, and ascertaining if the performance of the your application in the cloud platform is within agreed SLAs and your organisation’s ongoing business requirements.

In the current cloud environment, even standing still (i.e. keeping your service where it is) can cost money. The visibility provided by analytics takes some of the pain out of dynamically managing your service needs, whilst giving you the visibility to make informed decisions on capacity, performance and cost.

Are you clear for launch into the Cloud? posted on Sumerian analytics blog

Analytics timeline posted on Sumerian analytics blog

The diagram shows a simple multi-tiered e-commerce application:

Logs of most everything that happens are available for each of the main components of the system. If we can join these logs together we can trace each customer transaction through the overall system and produce an analysis which shows the end-to-end throughput of the system in business terms, i.e. what was delivered, what it took to do it and how quickly it happened. Having done this, we might, as a first step, plot some graphs to understand broadly what was happening in our business and our computer systems:

Interesting (if you like that sort of thing) – but not enough.

With a little bit of statistical analysis on the data we have collected, we can go further and establish the relationship between the demand from our business and the computer resources used in each system component and plot a graph of how our components scale against business need.

This gives us a predictive model of what happens to each system component  when business demand changes. If we have the tools then we can do this for every system component. Getting there.

The final technical step in this simple example is to use the predictive models we have built to show how much headroom we have in each component. In this example, we can see immediately that the current infrastructure provisioning is wildly unbalanced:

Who ordered those web-servers! Lets not go there. Rather, the GOOD NEWS is that we could more than double the number of transactions  processed  – HOWEVER the other components are way oversized – which in a topsy-turvy sort of  way is also GOOD NEWS, because we have options to either re-purpose the over-capacity or simply downsize. If this is done as part of an analysis prior to migrating to a cloud based service the savings achieved by resizing as part of the migration will be considerable. It seems clear that 15% target savings can easily be achieved in this system. We can say exactly what needs to be done to manage our costs – and how much that is worth is easy to calculate. The above example is not unusual in typical enterprise systems.

Now let’s zoom right back out and look how this could be extended over the whole of the IT estate. The following time-line shows how a large IT infrastructure can be analysed as a basis for a consolidation programme designed to reduce cost, deliver predictable service levels based upon business metrics and known headroom for growth.

We might decide to spread the consolidation plan over a period of time. If we compare this with our baseline costs we can see a positive NPV (net present value) over our planning horizon:

The savvy CIO who has been through their budget review with something like the above may well be tempted to start thinking how they could use the same approach elsewhere in the business. A good starting point might be a series of “did-you-knows?” such as:

• If we reduced the transaction times on our forex trading by 50% we would quadruple our revenue. Shall we build a business case for the investment?
• We will not achieve our planned 30% savings in Finance and HR unless we size our offshore SSC correctly – have we got the right capacity to deliver our business processes reliably?
• Low cost IT cloud provisioning is quick and cost effective when a business unit needs immediate capability for a new product launch – as long as it is decommissioned afterwards. Have we remembered to do that?

And so on. I am sure you get the idea. Analytics is not just for IT – it can deliver business benefit throughout the organisation – our example here is just that, an example, but one designed to show that Big Data really does contain valuable information – as times get harder it gets harder to find ways to improve your business. Companies that take the time to mine their data for the value will do better.

Big Data Analytics won’t fix everything – but in many cases it will certainly help. It’s about understanding complex operating relationships on a large scale, and it is certainly relevant to large businesses that want to lower costs without compromising their ability to deliver. Definitely, you should pay some attention to it. For starters, you might set yourself the target of reducing your cost per head for IT by $1000.

Big Data Analytics: Where’s the beef? – Part 2 posted on Sumerian analytics blog

You may have heard something like: “Analytics will monetise your business data.” A big claim – let’s see, by way of example, how this can sometimes be true. As is often the case, it is computer people who are mainly making the big claims for big data. So perhaps Enterprise IT is a good example to start with in establishing whether this hidden value can be revealed.

Consider your budget strategy this year: revenues look soft, the balance sheet is in reasonable shape, but nobody seems to feel very confident. The consensus is that it is a good time to play things safe and almost definitely hang on to the cash.

At the IT budget review meeting, the opinions from a range of perspectives might be something like:

• CEO (over-view): We need to innovate, spend less on running the shop, a bigger percentage on new capabilities to make us more efficient and win new customers.
• CFO (top-down):  Costs need to reduce, there is no appetite for capital investment and if things continue like this we’ll be reviewing our costs again at the end of the first quarter.
• COO (end-to-end): Services must be excellent, no possibility of failure and if business picks up, we need to scale up quickly. In the mean time, IT needs to take a smaller percentage of the operating budget.
• CIO (bottom-up): I wonder how I am going to do that?

Add into the mix that over the past three years: costs have been very tightly managed; suppliers have been squeezed; growth in demand for IT has absorbed most of the savings coming from technical improvements and a series of business transformation projects have re-directed monies that would have otherwise been spent on renewal of base capability. The challenge this year is compounded by the past three years. And on top of that – there is increased expectation of improvements in both flexibility and service levels – based on the consumer experience of IT.  It is getting harder to make savings.

Lets look at how Analytics might help – and by how much. For example, in the financial services industry this year you would be average if you spent 30% of your IT budget on a data centre and data network. Which in general might equate to around $7.5k per employee per annum.  If you could find a way to reduce that cost by 15%, then, say, for a 10,000 employee business, that is over $10m savings from an overall budget (opex + capex) of $250m. Obviously these are hypothetical numbers – though in many cases they will be realistic.  And while these are not astonishing savings – they are most certainly worth having – if not exceeding.

Making a list of ways to get at these putative savings throws up issues for the CIO, for example:

The ability to answer these types of issues will determine whether a business can achieve the savings it would like. One reason to be cheerful is that the basic data necessary to answer these questions are already owned by the IT team.  Most systems keep copious records of everything they do – but these logs are only rarely examined and aren’t that human friendly as a light read. Furthermore, each log is an island to itself – it doesn’t relate to the other logs easily. What is needed is a tool to bring these data together, join the dots and make the picture clear of how all this technology maps to business throughput. Then we can create the insight necessary to impact the business’s bottom-line – Big Data Analytics.

In part 2 of this blog, I’ll talk more about how this could be done in practical terms, showing how Big Data Analytics can be used to build a predictive model to inform a progressive consolidation strategy – and help achieve some dramatic cost savings.

Read part 2.

Big Data Analytics: Where’s the beef? – Part 1 posted on Sumerian analytics blog

My business card says “Senior Analyst” but that doesn’t tell you what it is I’m analysing – it could be anything from oil and gas prices to people’s dreams.  The last time I applied for car insurance I put my employment as “IT Consultant”, but that conjures up very different images of how I work from what is the reality (and also increased my premiums by £30 from the equally-inaccurate “Data Manager” I’d been the previous year).  I told someone recently that I worked in IT, and so was asked how to turn off the markup in MS Word.

However, there’s a new job title doing the rounds now that I feel describes me.  It’s perhaps not quite known widely enough yet for me to bring it out at the next dinner party I go to, or to use it to describe to my grandma what I do, but I’m hoping with a bit more weight thrown behind it, it might get there before too long.  The term is “Data Scientist”.

What does that say to you?  What images would it conjure if you met one at that dinner party?

For me, the “Data” part says that, while my job is generally around IT, the real focus of my job is the data that comes from that IT, not necessarily the IT itself.  Many of the techniques we use are transferable to any other area – in fact, I’ve done projects within Sumerian that had no infrastructure in sight at all.

While “Scientist” probably makes many people think of white coats and safety goggles, the important inference from it is the style of work – and particularly how it contrasts with more established roles.  I am not a Data Architect, who designs the data marts and schemas to hold the data; nor am I a Data Manager, who makes sure the data flows through the data mart properly.  I am a Scientist of Data – I experiment with the data – I study it and investigate the properties and relationships of the data.  I work to establish a deeper understanding of the data, to increase the sum total of knowledge of the data and to push out the boundaries of data exploration.

If that all sounds a little grandiose… well, maybe I am indulging in a little hyperbole.  However, I hope it does help to establish the difference between the work I do and that which my car insurance company thinks I do.

It’s important to stress that being a data scientist isn’t really a new invention – I’ve been doing this work my whole career, and plenty of other people have been doing the same sort of work a lot longer than I have.  In fact, the increasing prominence of data science began in the 90s when people started analysing the purchasing trends of their customers for marketing purposes, and that’s probably still the area where most data scientists work.  There’s a commonly quoted example about a supermarket that realised that people buying nappies had a higher tendency of buying beer too, so they moved the two items closer together to sell more.  Similar analysis drives the recommendations that many shopping websites will present you, based on what you’ve looked at or bought before.

But data scientists aren’t just about selling stuff to people. They can help today’s CIOs optimise IT environments too.

When tickets for the football at London 2012 went on sale last year, the sheer volume of people trying to buy them caused the website to all but seize up.  Using data science, we’ve carried out advanced analytics taking data on user volumes and system performance and used it to work out the capacity of similar eCommerce platforms – allowing organisations to provision extra capacity when they need it and avoid such outages.

But that’s just one example. It can be applied to all manner of IT challenges – from helping organisations to transition to a cloud-based environment, to improving application performance for driving improved revenue streams.

Data scientists are now influencing much of the world around you – from what adverts are shown in the ad breaks of your favourite show, to how much you pay for your summer holiday,  to how well the website you’re using is performing.  However, the term is only now starting to crystallise around this group of people and give them something of an identity.  We’re probably still a fair way off the founding of an Institute for Data Scientists, or a Data Scientists Union, or even just getting our job role on the car insurance websites, but we’re here, and now we have a name.

Related: Sumerian Expert Analytical Services

The rise of the data scientist posted on Sumerian analytics blog

Last week on my Twitter feed there was a link to an interesting article by Pete Warden arguing that the ancient Sumerians invented data, recording information on stone tablets, some of which survive today from the 4th millenium BC.  Now, as a 21st century Sumerian, I find this interesting because Pete likens the practice of the ancient Sumerians to our modern data science concepts, for example:

• Data is power: The recording of transactions can be used to demonstrate the truth and therefore assert power through knowledge.
• Data provenance: Transparency is required to prove the source and therefore validity of data.
• Data integrity: Ensuring the content of data is valid and has not been altered.
• Data interpretation: Human dynamics influence the understanding of the final results.

At Sumerian, we have yet to receive any requests to analyse data from stone tablets, but we have recently received samples supplied on a floppy disk which now seems like an ancient technology.  I wonder if, and in what form, the mountains of data from Twitter will be available in 6,000 years time.

Related: About Sumerian Analytics

Sumerians and the value of data posted on Sumerian analytics blog

Latency is a continual challenge in trading systems. In High Frequency Trading (HFT) systems, the challenge is immediate and obvious: if your order isn’t in first, you won’t hit the liquidity you’re after.

In other systems, it might not be so obvious, but it’s still an issue. For example, if you’re streaming FX quotes out to OTC venues, any latency you introduce increases the chances of a subsequent order being rejected due to changed market conditions; do this on a regular basis and your counterparties will quickly become wary of dealing with you, resulting in lost order flow.

Given this challenge, it’s not surprising that pretty much everyone in the market is working hard to reduce latency in key systems. This usually starts with an effort to measure and benchmark existing latency. After all, as the old adage goes, if you can’t measure it, you can’t manage it. And here lies the first challenge – in most trading systems, latency isn’t something that happens in one place; rather, it’s the time difference between information arriving from a source, and a resulting order/quote/trade being sent to a destination. Depending on the complexity of the system, there may be dozens of infrastructure and application components between these two points. If you only measure the end-to-end latency, then you’ll know whether it’s good or bad, but where do you take action to improve it? Likewise, if you only focus on one point (your market data source, or exchange connections, for example), how do you know its latency contribution is a significant component of the end-to-end trip?

So, if the objective of the process is to make improvements in your latency, then you need to be able to quantify the contribution of each of the components in the path. Now, there are numerous ways to do this. You can leverage existing log files to track flow across multiple components (assuming you can address time synchronisation challenges); you can capture packet traces and use these in a similar way; you can purchase one of the many dedicated latency monitoring solutions on the market; or you can use some combination of all of these. The best approach for you depends on a whole number of variables, and could easily fill another full article. Let’s assume you’ve chosen and implemented your approach, and you’re now measuring your latency, and how much each system is contributing.

Now the fun really begins …

Because measuring latency only tells you the what and the when. What is my latency now? What was it at 11am on Friday? It doesn’t tell you the why or the what if. Why is it higher than usual? What will it be when market volumes double, or when you add new customers? In order to truly manage your latency, you need to do more than just measure it – you need to model and analyse its relationship with other things that are happening in your environment: market data volumes, order volumes, network throughput, infrastructure utilisation and any other component that could affect it.

This is where IT Analytics comes to the fore. By bringing together the latency data, plus business volume data, plus infrastructure metrics into a single, large, normalised data set, you gain the ability to understand the relationships between them. You can quantify the impact of business volumes on latency. You can find the components whose latency is most sensitive to volume and quantify the improvement achievable by re-engineering them. You can identify the infrastructure components where capacity limitations are causing latency spikes. And because these Big Data models are re-usable, you can do all this repeatably and consistently – so you can quickly see evidence to validate any improvements you make.

This isn’t a pipe-dream. Yes, there are issues with data quality; yes, the data sets are large and the normalisations can be non-trivial; yes, finding the relationships requires specialised statistical techniques. But advances in Big Data in the past few years make all of this achievable – as an example, Sumerian helped one customer reduce the end-to-end latency in a key FX flow by 75% by using exactly this approach – and did it in less than eight weeks.

There are big wins to be had from applying IT Analytics to data you already have. Organisations who take advantage of it stand to gain that all-important competitive edge, achieving systematic, ongoing reductions in latency.

Related: Sumerian Latency Analytics

Eliminating latency with Analytics posted on Sumerian analytics blog

In 1991, I won the creative writing prize in junior school with a phonological delight that started: “Clouds, clustered in clumps, like balls of cotton wool’. It was a Homeresque masterpiece of about a hundred stanzas whose chief literary technique was unrelenting alliteration (a favourite employ of the post-kindergarten bespeckled set). Anyway, I won it none the less and proudly spent my £5 prize on a book about computers, the rest as they say is history.

Now, 20 years later, I’m back staring at clouds. 2011 has been a year full of clouds; for one thing, I live (for the most part) in Scotland, and as anyone who’s visited will testify, the Scottish are a world authority on clouds, or more precisely rain. Edinburgh is probably the only city in the world where offices are legally required to carry liferafts and we all have floatation devices stored in a little pocket under our desks,such is the omnipresence of the chin-dripping downpour.

Aside from their hand in the prevention of anything that would generally be recognised as a summer in Scotland,it’s been a big year for clouds – in fact, the world has gone positively cloud mad. Brand “cloud” has been slapped onto everything – from soft drinks, to games characters, to dance festivals and even eggs! – in an attempt at making it look like the hottest thing since Cloudia Schiffer, and herein we fall tenuously towards the point of this monologue.

At the moment it seems like every IT vendor and his tennis partner is talking about, and trying to sell “the cloud”. And I can see why. It’s a perfect nexus of tangible marketing opportunity, available technology and actual need. It’s also because for the first time, consumer grade tech is actually better than that being adopted by business. Previously it was businesses who walked forrader in the technology promenade, but in Cloudville it’s really consumers who’re leading the way – they’ve got Flickr, Facebook, Google Docs and ever more by the day, whereas for most large enterprise businesses “the cloud” still means internal server virtualisation.

This unprecedented level of technological familiarity in the retail market does have its upsides. For one thing, it makes the previously impenetrable life of an IT exec a little more accessible. As anyone with a mother will know, working in IT and being able to explain what you actually do for a living is almost impossible, so having a topic broadly understood by home users is dinner party manna. This accessibility, and henceforth marketability has led to a plethora of IT vendors slapping the word “cloud” on their product, whether it actually has anything to do with flexible, remote, agnostic processing or not.

Cloud technology has been around in various guises for 30 years or more, but the current typical business implementation (private datacentre flavour) is really just an extension of our previous decade’s virtualisation movement, with some added flexibility and perhaps a stab at recharge and active capacity management. But there’s actually much more to it than that. When implemented appropriately (and if necessary, multifariously) cloud technology can be highly cost effective while actually improving end user experience. The key theme here is appropriate deployment.

The first step is to really understand application and service demand at a client level, then mix in the performance, security and other business considerations before testing the business case and only then considering implementation options. This is an area where one size definitely doesn’t fit all and so working in tandem with a well defined service catalogue is essential.

Throughout each of these stages access to insight based on real data is key. Without an analytic approach you might find yourself riding a public Nimbus rather than a hybrid Cirrus, and who knows where that might end up. In summary, without the right data and skills to make sense of it, it can be pretty hard to get a firm footing on a cloud, and if you believe all the marketing hype you might end up with nothing but vapourware.

Related: Sumerian Analytics for Enterprise Cloud

Clouds, clustered in clumps, like balls of cotton wool posted on Sumerian analytics blog

It’s fair to say that during much of this period, the corporate IT equipment we used was far better than anything we had at home, if we had anything at all beyond a calculator. That gradually began to change and a look around many homes now shows how sophisticated our personal devices have become. Tablets with app stores, Android devices and iPhones, dual boot Windows/Mac laptops, not to mention full-on PC media centres.

As consumers, the services and applications we use have also become very powerful. Web 2.0 technology allows us to do our banking and shopping with a few clicks. E-books and newspapers are automatically delivered to our desktops. Search engines allow us to track down information, no matter how obscure, in milliseconds. Services are now being hosted in clouds,allowing us to seamlessly move between devices whilst accessing our personal data.

Meanwhile,back in the office, things seem to have been proceeding at a different pace. Lots of time and money has been spent upgrading operating systems, installing SANs and faster networks, moving from client-server to web based technologies, but for the end-user have things changed that radically? Most of us will have been going through laptop upgrades every couple of years and we may have picked up a corporate mobile or smartphone along the way. However many frustrations remain: slow laptops, difficulty in tracking down corporate data, having to endure time consuming OS updates and patch installations. The consumer IT world has eclipsed the corporate one for many of us, in terms of both devices and the overall end user experience.

Many organisations are waking up to this fact however; CIOs realise that the one size fits all approach is no longer relevant. Joining an organisation should no longer mean you get issued an identikit laptop or desktop and possibly a BlackBerry if you are senior enough. Thin clients can work for office based staff, Mac or Wintel laptops for knowledge workers, tablet style devices are ideal for roaming staff that just need to access eMail, review documents and use specific apps whilst travelling.

Technologies now exist to support this consumer style choice in the corporate world, such as desktop virtualization, private and public clouds, application virtualization, enhanced mobile coverage and better security on commodity devices. Some of these technologies, as well as supporting a range of end user devices, can also help alleviate some of the common frustrations we experience in the corporate environment. For example, virtualised desktops held in a data centre mean that patch and OS updates can be performed whilst the user is offline, instead of when they are trying to start up or shut down their laptop.

To make this leap, organisations need to determine what their users need to do their jobs and then match that to the various technology options on offer (not the other way around). Sumerian has worked with clients in exactly this scenario, analysing how users behave, understanding their working patterns and then matching this to technology options.

Techniques such as data mining and clustering, long used by supermarkets and e-commerce companies to analyse consumer behaviour and buying patterns, can be used to understand how staff use IT services and technology. We can uncover hidden groups of users, each with different behaviours and needs, allowing us to determine how new technologies can help them in future.

While there are technical and organisational challenges to be overcome, the benefits are numerous. More productive and flexible staff, cost reduction through more efficient desktop management and recognition as a leader in technology adoption, to name a few. We may even avoid the visit to the physio every six months as we don’t have to carry our laptop everywhere we go.

Related:
Sumerian Expert Analytical Services
Best practices to maximise end user services

Desktop revolution #9 posted on Sumerian analytics blog

As a bit of an avid gamer, I’ve been affected by the Sony Playstation Network hack…

While the reason for the hack hasn’t been shared yet, it’s clear that it’s had a massive impact on the user base. Password and personal details have been compromised. Credit card details, although encrypted, have been snatched. Not only that, the hack went beyond only the PSN – an extra 25 million online service users, from massively multiplayer online role paying game (MMORPG) players to casual Facebook gamers, have had their details stolen. It’s the biggest hack in history.

Recently, Amazon’s AWS service experienced a significant outage – having an affect on many well known organisations and businesses. A good summary of some of the opinions around the AWS outage can be found here.

All in – it’s been a bad couple of weeks for cloud. If I were a CIO contemplating a cloud migration, these recent occurrences would give me pause for thought. And rightfully so.

Can you trust a 3rd party with your data? What are the implications of a significant outage?

You might run your overnight batch in the cloud… But what happens when it’s not available? What happens if it fails to complete? What happens when you can’t meet your obligations to the rest of the business and the reason for that has been out with your control…

These recent outages, and there will be more to come, demonstrate the importance of ensuring you understand the operational risks of various cloud models and plan accordingly for mitigation. While many companies will have designed for outages,this may have been the first real test of such a scenario.

Cloud computing is still an emerging technology. The industry is still cutting its teeth… I’d say it’s in the “terrible twos” stage. It’s pushing the boundaries and finding its place. It’s not going to go away and,if anything, these recent events are going to make it more robust, reliable and trustworthy. The experiences and lessons learned from these events will lead to stronger and more resilient platforms.

In the meantime, private cloud will continue gain popularity. Offering the agility of cloud but still within the control of the business is an attractive option. The responsibility for availability lies entirely with the organisation and costs can be continued to controlled centrally. Over time, these private clouds may dissipate – but not anytime soon, and not with any more of these significant events and outages.

Related: Sumerian Analytics for Enterprise Cloud

When clouds turn nasty posted on Sumerian analytics blog