This rapidly growing organisation wanted to move more services to the cloud as its on-premise servers were reaching end of life, but first needed to rationalise its existing cloud services, which had grown haphazardly and were expensive and difficult to manage. After analysing the problem Fordway implemented monitoring and reporting to stop costs escalating and developed plans to transform services into a smaller, more manageable cloud to improve management and security and reduce costs.
- Cost reduction
- Improved governance and control
- Departmental cross-charging and accountability
- Optimised architecture making best use of services
- Easier to support and manage
Treating cloud as if it was in-house capacity.
When we first started working with this organisation it could have been a case study for the perils of treating cloud as if it was in-house capacity. The organisation had grown rapidly but its in-house IT infrastructure was unable to scale to handle demands. Each department had a high degree of autonomy and required its own projects and specialist applications. Internal provisioning took too long, so the IT department introduced public cloud as a panacea or cure-all for the challenges it faced. Each department was given a separate Azure tenancy to develop and run its own applications, business intelligence and new projects as it saw fit, while central IT provided assistance, support, security and management. However, there was no guidance or planning – something which we often see in organisations taking their first tentative steps in public cloud.
When Fordway was appointed the organisation’s twelve tenancies had created governance, risk and security challenges, as well as increased costs and infrastructure sprawl. Meanwhile the lack of planning meant that the problems with capacity and end of life equipment had not gone away as hoped, and the organisation had not achieved the expected efficiencies of cloud.
Fordway was able to lock down the sprawl, providing monitoring and KPI reporting to enable actionable management information and to baseline the organisation’s experience. We then developed plans to transform their services to a far smaller and more manageable cloud footprint to make day-to-day administration and security considerably simpler and start to pull back costs.
Unplanned subscriptions created a complex architecture
The organisation had begun moving applications into Microsoft Azure around four years ago. The number of Azure subscriptions grew organically as users requested more services, and as each new subscription was added it was connected to the others using VNet peering and gateways to enable traffic between them.
There had been no high-level design or optimisation, resulting in a spider’s web of connectivity. Costs had also increased significantly because inter-tenancy traffic is metered, and therefore chargeable, which had not been anticipated. The large number of Azure subscriptions and the gateways and traffic between them cost several thousand pounds per month, which was much higher than the volume of data and traffic merited. The incumbent managed service provider had been too busy problem-solving to address the issue, so the problem continued to grow.
With no standard template for new instances, the organisation had a mix of instance types and sizes, configurations and costs, which made bills more difficult to analyse. There had not been any proactive instance management, meaning it was using older or retired instance types with lower performance and higher costs than more recently released instances. It was also using different storage types and did not have any consolidated storage accounts.
Treating cloud instances like server capacity increased costs and made support difficult
The problems had arisen because the organisation was buying cloud as if it was in-house server space. Each time someone asked the IT department for more capacity, instead of building a virtual server they were given a new Azure instance. However, while an in-house virtual server uses on-premise capacity, so does not add to overall spending, each new Azure instance incurs a cost.
The problem was exacerbated by the way new instances were obtained; the cloud service provider had set up a portal, so buying a new instance was a simple matter of point and click, with no consideration of the most appropriate sizing, service or functionality. As a result the organisation was not benefitting from the cloud’s flexibility and scalability by optimising the type of services chosen and only paying for capacity as and when needed.
The complex configuration also made it difficult to provide support, as every time an issue arose it took the service team a long time to find out where the problem originated. This meant issues took longer to resolve, causing problems for users and increasing support costs.
When Fordway reviewed the situation, we identified a further problem: tenancies had been set up in two separate Azure locations (Availability Zones): UK South and Western Europe. The UK tenancy was used for all the active applications. No-one in the organisation knew why the second tenancy had been set up or what was on it, but it incurred costs every month.
With some on-premise server capacity already at end of life and the remainder reaching end-of-life in a few months, addressing the issue had become a priority. The services currently supported in-house needed to be moved to the cloud, but the organisation did not want to add any further capacity until the existing network had been untangled.
Effective planning reduced complexity and costs
Fordway carried out an extensive analysis of the business and concluded that the Azure services could be consolidated into four subscriptions on a single tenancy. These could all be managed through the new portal: one for the core platform, to be used for business intelligence, and one each for pre-production, production and testing. This work would also allow migration to the latest Azure instance types, as it created a framework enabling migration to be optimised against workload, data consolidation and reclaiming storage reducing costs.
With this high-level design in place, a detailed plan of IaaS, PaaS and SaaS transformation models, capacity planning and enabling high availability will improve performance and costs to be optimised. This consolidation and optimisation will result in savings of several thousand pounds per month as well as simplifying management, improving security, reducing operational overheads and removing bottlenecks.
Azure’s resource tagging ability, which provides highly granular and transparent information on capacity use, can then be employed to track and categorise costs, and to recharge service consumption back to individual departments. If a department requests additional capability, it will become straightforward to identify whether they already have suitable capacity and permission to establish new services. This will create a new culture of accountability, as well as saving money from reducing the number of online services with no workloads. The reduced complexity, in turn, will make support faster and more efficient. The organisation has already recognised the need to manage its use of cloud more closely and is currently recruiting a director of service delivery.
Find out more about Fordway cloud optimisation services.