It’s funny when you think about it, cloud infrastructure provides us with the capacity to power and host the most cutting edge computing technology, allowing the latest scientific and technological innovations to go above-and-beyond, but the way in which AWS is managed and how it interfaces with developers and companies is really pretty dated.
I mean “tagging” resources? Isn’t that what we were doing 10, 20 years ago? It’s not dynamic, responsive or even automated. It’s pretty much data entry. No wonder developers get frustrated when they need to spend time on this task.
Traditional tagging policies involve creating an excel-like sheets within the AWS console. Every resource employed (ie. the AWS instances ordered by developers) should get given one or several “tags” so there is a way for them to be tracked and identified.
Resources without a tag are just seen as a sort of machine serial number. Tags, therefore have been needed to indicate more information about what a particular resource is used for, by attributing such information as country, client or developer. This means one can see which resources are being used and how much they cost… and even on which project and by whom, it creates a way of tracking where the money is being spent. From these tags you can sort out the content of billing files, to create a graph table, or pie chart to identify where cost is being dedicated, according to the tags.
Yet, the problem with tagging isn’t just that it lacks sex appeal, in fact like many analog based tasks, it's prone to error and there is a complete lack of flexibility. Small input errors can equal big analysis errors. So while a lot of time consuming frustration is being put into tagging AWS resources and the like, the results are at best an accurate record of the past —which is arguably not especially useful, because it’s already happened—and at worst completely misleading.
The input errors that can lead to misreading or wrong data for cost tracking, can come in several varieties, firstly a simple typo in a tag means the corresponding resources won’t be attributed to the right (or any) group. Even just using a different format creates this same problem. If someone rights RandD instead of R&D, all those resources won’t be included in the R&D category when it comes to tracking costs. And how about when a new recruit gets landed with the tagging gig (and yes this is often the case), he doesn’t use the same system or terms that conform to your company's way of doing things.
Non-scaleable = non-viable
Things get even more complicated when companies get bigger or a new strategy from management means seeing more detail on cloud spend and more tags are required to visualize and categorise differently. Management might suddenly want to see geographical regions (not the cloud datacenters, the regions of your financial P&L for example) as well as which resources are dedicated to which clients. Or what if a resource is dedicated to another project mid-way? Who’s going to go back and retag all previously tagged machines? It’s a time-consuming manual process that doesn’t lend itself to scaling up. As a results it's used poorly and almost always inaccurate or obsolete
So we decided to get rid of traditional tagging
So while some cloud infrastructure management companies are talking about the best tagging strategies to adopt. The fact is, tagging, in it’s current form, is a non-optimal way of doing things. That’s why we completely got rid of the need for tags, meaning we allocate resources to projects, and like in the “real world” the projects depend on teams or users. This is usually all that’s needed and is already a massive improvement. A change of ownership or org chart? All cascading budgets, past and projected spends will be recalculated, accurately and in real-time. Thus, eliminating the need for tags and the dreaded tagging task. We can still provide tags for resources if that’s what works for you and even assign them across projects and teams.Project-centered not slave-to-the-tag
You and your teams can forget about complex resource—and often obsolete— tagging policies. We are project-centric, we allocate resources to projects, instead of the other way around. It’s simple really. Each project has a unique personalized key, per user, so it’s easy to see who is doing what, when and with how much. Tracking a project is automatic, so it doesn’t require any work.
Want to try it for yourself? Sign up for a free trial and see how life without AWS tags saves money, work and time.