By Dan Roncadin – Chief Consultant

While advising companies on their AWS environments, a few core issues come up again and again. These issues present real challenges for organizations moving to broader adoption and aren’t easily remedied with a quick fix.

1. Cost Management

Moving away from upfront capital expenditures to on-demand usage-based pricing is a primary driver of many clients’ cloud adoption strategy. It’s easy at first. The bills start small, a few hundred or thousand dollars a month. Low enough for someone to expense on a purchasing card. Soon they get bigger, 10’s of thousands per month, then 100’s of thousands. At this point, it is a serious expense and starts raising questions about efficiency and forecasting. It’s very clear what the market price of a traditional server is, same with storage and bandwidth, but with AWS it’s all about how you use the services. If you don’t have a very deep understanding of how the costing models work and what options are available for tuning, you are at a disadvantage. You could be achieving the same levels of performance and availability for less money.

Once you’re spending six figures monthly there are likely multiple consumers of AWS within your organization, and you need to ensure that costs are tracked back to the users appropriately and managed. There can be multiple billing accounts in use, and this can be a lot of data to deal with. Some helpful tools are available in this space that can make tracking costs easier, but properly optimizing takes a more strategic approach. Many tools only help with simple things like suggesting reserved instances. Often the real optimization opportunities come from going beyond simple reporting to analyzing the architecture and design of the environment to ensure it is using the right services in the right ways.

I’ve worked with a number of large clients analyzing AWS spend and often find significant savings opportunities (>30%), especially in those spending over $100K/month.

2. Translating Organizational Knowledge of Traditional Infrastructure into Cloud

Many clients I work with have great teams of intelligent and experienced technical people, who unfortunately have little or no experience with the cloud. It’s a challenging market to find people who have this experience, so this is usually why they’re talking to Taos about it. Although there are similarities between traditional infrastructure and cloud offerings (“instances” are like servers for instance), there are enough differences that aren’t intuitive to make a difficult transition. Many other derivative issues that end up plaguing environments, such as problems with Security, Performance, and Availability are rooted in a lack of knowledge and experience with AWS. Knowing how to design, deploy, and operate a traditional datacenter doesn’t adequately prepare you for AWS.

3. Keeping Up With The Pace Of Change

Amazon releases new features daily, and they rarely publish details of what’s coming up in the future. That alone makes keeping up with the pace of change a serious concern for most clients. There is no place for a waterfall approach that only does design and architecture upfront then moves on to deployment and operations. All parts of the environment design need to be re-evaluated on a continual basis every time one of the constituent services at AWS makes a change. Many operations organizations expect fundamental infrastructure components to be relatively slow to change, and aren’t able to keep up. Consequently, many environments who moved to AWS more than a year ago but haven’t made significant changes since being deployed, are now not optimized from either a cost or operations perspective. Part of the key to operating with AWS successfully is designing to expect change and be able to react to it quickly.