We recently had the chance to speak to Peter Duffy, CTO of Sumerian, about the past, present and future of capacity planning and the importance for companies to stay on top of their IT estate.
The full interview can be found below.
- Firstly, give us a brief background of Sumerian and what capacity planning is all about.
Sumerian is an Edinburgh-based capacity planning company that applies advanced IT Operational Analytics to enable our clients to improve the overall performance and efficiency of their operational IT. Our clients range across a variety of industries including pharmaceuticals, finance and retail, however our product appeals to all organisations that benefit from effective management of their entire IT estate.
Capacity planning is all about providing CIOs and IT managers an in-depth understanding of their IT estate, giving them insight into their headroom for future growth, and helping them to avoid capacity-related incidents and save money. It’s the best way to go about IT consolidation and optimisation and de-risking planned changes to IT estates.
Sumerian helps businesses unlock the potential of their data in order to predict their IT future. Utilising the data from server log files and system management tooling, our Capacity Planner CPaaS enables businesses to run powerful predictive analytics and ‘what if’ scenario modelling on their IT infrastructure.
- How important is it for companies to stay on top of their ‘IT estate’ and is this becoming harder to do?
Most enterprises today depend on a reliable and performant IT infrastructure - it’s customers and business success rely on it. Consequently, it’s critical that IT teams ensure that potential performance impacting incidents are proactively avoided. Additionally, every company is run on a budget, albeit some more tightly than others. Being able to rightsize their IT estates, companies can ensure that they are not over-spending on their IT resources.
It is certainly becoming harder for CIOs and IT managers to stay on top of their IT estate, and there are several reasons for this. The dynamic nature of demand - due to the ever increasing amount of devices being employed by users, data is coming in from anywhere and everywhere. As hard-drives fill up faster, companies need to ensure that they understand the rate at which their IT resources are used up, so that they can predict their future capacity needs, and thus provide extra capacity when needed.
Companies are also increasingly adopting the latest technologies, including Virtual Machines (VMs) and cloud, in addition to their on-premise hardware – creating more complexity in the IT environment. For example, many IT teams lack the foresight to remove VMs from their systems after discontinuing use, this can create chaos when they all start up again after a power outage.
Proper capacity planning enables businesses to understand which VMs that they have on their systems, and de-risk their provisioning. Amongst this complexity, companies find it difficult to understand accurately how their IT resources are being consumed, and whether or not they are investing efficiently.
- What are the main ways capacity planning can help businesses?
Our capacity planning helps those responsible for IT infrastructure to gain a new level of control. By providing accurate visibility of potential future capacity events and enabling a longer planning horizon, businesses can take proactive action confidently.
Through our ‘what if’ scenario modelling, businesses are able to stay ahead of the game, accurately simulating planned transformations and quickly seeing the impact of planned changes to their IT estates, balancing risk and cost, therefore driving higher consolidation ratios, redistributing work-loads and optimising their estates across both physical and virtual platforms.
- Are there certain industries where capacity planning can have more of an effect than others?
There are definitely certain industries where not having a stringent capacity planning process in place can have even greater repercussions than others, such as financial services and pharmaceutical. These tend to be sectors that receive an intense level of focus from both the public and government/regulatory bodies regarding their IT infrastructure and operations management.
For example, a bank’s customers put a lot of trust in them to be able to access their money whenever and wherever they need it. Any disruptions to their services and national headlines are made, and millions of pounds in fines are received. On top of this, untold damage is done to their reputation as a financial services provider in what is a highly competitive industry.
This is why banks are under strict compliance regulations put in place by governments. Not only is this a huge incentive for banks to make use of capacity planning, but due to the sheer size of a bank’s IT estate, they stand to save a lot of money through consolidation and rightsizing, as well as planning properly for any changes to their estate.
- Can you give any examples where capacity planning has significantly improved a company's performance?
We are currently unable to name our clients for you, however I can give you the example of a global bank that wanted to improve the management of its operational risk and reduce IT operating costs by $3 million over two years.
The bank had already used Sumerian’s predictive analytics to drive reductions in FX trading application latency and to improve batch processing times. By using Sumerian’s CPaaS the bank is able to decide whether current levels of operating risk are acceptable or not, and determine precisely when and where additional investment is required across its estate.
Thanks to Sumerian’s CPaaS, the bank avoided 300 service outages in critical systems last year alone, and avoided $1.2M through right-sizing.
- A lot of what you do with clients is based on predictive analysis. What are the risks involved and how are they mitigated?
It's obviously not possible to be 100 per cent accurate when you are predicting the future, but there are a number of things you can do to maximise accuracy, to give users an indication of confidence in predictions, and to err on the side of caution.
First off, when we are finding patterns and trends we give more weight to recent data than to older data. That way if behaviour starts to change you detect that change more quickly.
After you have created a model (which is basically a mathematical equation that uses information from the past to predict the future) you can have your algorithms test that model against different data. If the model predicts values that are 'close enough' to the data then you can decide if it's a good model, and use it for predictive purposes. You can quantify this 'close enough' criteria by calculating the difference between actual and predicted values - so we do this and only use a model if it passes a particular quality threshold.
And then finally, when you show predicted values to users, you can put confidence intervals around them - so you can say "we predict utilisation will reach 90 per cent next Thursday, and we're 95 per cent certain it will be between 87 per cent and 93 per cent", and that way users can decide what to do with that information.
By doing all of that (and a few other tricks) you can make sure that you're not overwhelming users with false positives, but at the same time you're not missing major issues.
- How has the cloud helped or hindered the process?
Cloud providers use a mixture of terminology to describe their ‘flavours’ (e.g. small, medium or large) of compute and storage offerings. Much like clothes shopping, a ‘large’ from one vendor might be a ‘medium’ from another. This means it is often tricky and time-consuming to meaningfully compare one to the other. Capacity planning allows organisations to easily model these different flavours to rapidly understand how they compare to each other, and what they will need to support particular workloads.
Building an accurate plan for migrating workloads as part of a managed process (and being able to clearly communicate these to all stakeholders) is key to making a success of migrating to the cloud. Sumerian Capacity Planner allows organisations to easily create application or service specific views to clearly understand the demand generated by the components within the application/ service. Using that accurate data, various migration scenarios can then be quickly created and tested to find the optimal migration plan.
Cloud isn’t necessarily about reducing costs, it’s perhaps more about increasing agility, about having that capacity on demand. This, however, introduces new problems. Application teams, no longer reliant on centralised IT capacity, can buy what they need, when they need it. Users will always find a way to access the IT resources they need.
As you can imagine, this can lead to spiralling IT costs. Suddenly that ‘cost benefit’ of adopting Infrastructure as a Service (IaaS) vanishes and the business has no way of understanding and controlling their IT spend. The point being that, for organisations using cloud technologies, understanding the business requirements is vital in managing IT costs and requires a strong and visionary IT department.
- What future trends do you expect to see in the industry over the coming months?
While capacity management is an old discipline, it is currently a resurgent trend and is coming back into focus. The continued growth in the adoption of cloud computing technologies such as virtualization and ever-rising levels of business demand have contributed to an increasingly complex IT environment.
On top of this, businesses are having to look at different ways of dealing with capacity issues other than just the ‘adding more tin’ approach, as they look to streamline their expenditure. The insight provided by applying new capacity planning approaches such as predictive analytics and accurate scenario modelling enables businesses to make confident decisions concerning their IT estates without having to rely on guess-work.
We also expect to see continued growth towards everything ‘as a Service’, as organisations exponentially increase global spending on hyper converged infrastructure and seek to optimise their IT environments to meet workload needs. Following this trend, capacity planning is now also being delivered as an on-demand, cloud-based service – CPaaS, which gives instant visibility and accurate forecasting across organisations’ entire IT estates.
All in all, the growing complexity of today’s IT environments and the ever increasing demands of you and I – the end users - presents significant challenges to IT managers as they attempt to balance growth, investment and risk.
Gathering intelligence on each of these aspects is as important as ever, and capacity planning will be high on the agenda as businesses seek to future-proof their businesses and de-risk their IT operations.