dsp are very proud to be partnering with a well-known brand to move them into the Public Cloud for their Enterprise systems.
Our clients’ Virtualised Microsoft SQL Server Infrastructure hardware was coming to an end of its life. This included dedicated SAN’s, management tools and a pool of SQL Servers with over 120 databases. These served Sharepoint, CRM, Accounting and line of business applications critical to the organisation.
dsp was engaged to help determine if Cloud (either Public, Hybrid or Private) was right for their production systems. Being their core systems, it was absolutely critical that this decision was the right one and no matter how much cost could be saved, it was the services to the organisation that was priority. A detriment to service wasn’t a trade-off they were willing to make to save cost.
Therefore, there were drivers and concerns which needed to be weighed up, business case to be drawn up and advice and technical clarification was needed on the as-is and the to-be Data Platform.
Drivers for Cloud
Cost – indicative initial costs definitely showed a real saving could be had
Budgeting – Not having to present the FD with surprise bills throughout the Fiscal
Manageability – Timely and costly onsite infrastructure meant an increase in headcount if it stayed on premise
Scalability – Predictable and unpredictable growth was difficult to budget for and capacity plan for.
Blockers for the Cloud
Performance – Will the users notice the difference? Taking everything from compute to latency into account
Compliance – Where will my data be? Are there any regulatory limitations?
Technical Flexibility – How will my current and future initiatives fit into this if I move my Data Platform to the cloud? Integration with O365? Business Intelligence tools? Legacy Applications?
Cost – Are the cost savings real or pumped up by marketing people? How will this help me in future?
Journey to the Azure Cloud Platform for MS SQL
From the initial ‘I’d like to investigate moving to the Cloud’ statement, there were certain steps we took to determine if Cloud was the right move.
Step 1 – Indicative Costs – were presented for basic compute, storage and network; as well as Professional Services and Proof of Concept Costs. This was to determine if the journey was worth taking; and determine the cost of the journey and the savings when they got there.
Step 2 – Requirements gathering – dsp needed to understand the environment, specific applications, how busy the SQL environment was and any known problems that would scupper a POC or Cloud feasibility, Disaster Recovery and High Availability options, future plans and initiatives, and required Cost models (budget release and allocation? Capital versus OPEX?)
Step 3 – Inviting Cloud Providers to Bid – A healthy mix of high end IaaS Providers, public and boutique Cloud providers each with their own benefits and value proposition.
Step 4 – Choosing the Platform and Shortlist bidders – dsp consulted on the specifics of the Data Platform configuration. Which BI tools needed to remain onsite, what application servers need to be in the cloud and not be in the cloud, and what databases/servers would be problematic in each environment. The client then shortlisted the IaaS bidders taking our advice into consideration
Step 5 – Proof of Concept – A critical stage. dsp helped determine the criteria by mapping the databases and documenting key traits to look out for during the POC. Whilst we provide POC services, it is essentially a piece of work that cannot (and should not) be outsourced completely to a 3rd party (dsp included). It is your users that will be affected and you know your business better than we do! So with the POC we set the guidelines and test criteria and worked with our client in an advisory capacity only. We did help ensure a fair and equal measurement of each bidder. This included load and performance testing and user testing in the live environment. Latency, supplier response, ease of use and technical flexibility were just some of the criteria set.
Step 6 – Final Mapping and Analysis – Once the POC’s were complete, dsp were commissioned to ‘rubberstamp’ the final design, performance, configuration and migration plan. This is sometimes done before the POC, but I suppose it didn’t make a difference to the client in the end… This took a few weeks. We had to understand and compliancy implications with the backup solution and calculate the IO costs as well as data extraction costs for the BI elements that were staying onsite. This could have been the difference between the projected cost and 60k difference in the first year if this exercise wasn’t carried out properly!
Step 7 Migration and Go-Live – Ensuring a smooth migration following the plans drawn up in Step 6. Migrating to the cloud is relatively straight forward, however some considerations around running the two environments in parallel need thinking about; and ensuring all the little infrastructure and BI/Application bits are understood and working at once it’s live.
Step 8 Support and on-going Management – dsp provide the SQL Database Administration on behalf of this client with healthy relationships with the client’s application team. The management of the Azure platform is also included. Effectively, the end result meant that the Infrastructure, Data Platform are on a true ‘utility’ based model. They need more compute the environment scales accordingly with costs already predefined. dsp are effectively now an extension to the client’s Data Platform and Infrastructure team.
So there we have it, not all companies will have the same steps and some will be a bit more cowboy’ish about it or not go in as much detail. Others will look to their existing IaaS providers to provide everything.
The key value dsp brings is that we’re on the client’s side and we look specifically at the Data Platform – which effectively dictates the infrastructure requirements and the application tier. Our client trusts dsp to be the dynamic, responsive, trustworthy strategic partner, working alongside them in this brave, cloudy new IT world we find ourselves in!