Join 16,500 others and follow Sean Hull on twitter @hullsean.
1. Complexity Is Growing
Despite automation & the mass migration to the cloud, or perhaps because of it, complexity continues to grow. Back in the dot com era a typical infrastructure included a load balancer, a couple web servers, one oracle database, and that was pretty much it.
Now that has multiplied. Pile on top of that three to five more webservers, a search server, a page cache, an object cache, one or more slave databases and more. You may have a utility server with jenkins for continuous automation, monitoring applications like nagios and cacti, your source code repository and perhaps configuration management like Puppet or Chef.
That’s not only more moving parts, it’s a wider swath of skills and technologies to understand. That’s one reason Generalists Are Better At Scaling The Web.
Also: Are SQL Databases Dead?
2. Developer Mandate: Features
The pressure to build features that can directly be monetized is obvious. Startups especially have the pressure to grow fast and grow now. So security, technical debt, and scalability often take a back seat. What’s more in small scrappy and lean startups, ops sometimes falls on the shoulders of one competent but overworked developer.
Related: Why Oracle Won’t Kill MySQL
3. Startups Growing Pains
With hyper growth, startups can go from 100 customers to millions overnight. That kind of popularity is a good problem to have. But if your app hits a wall and suddenly falls over, everyone is scrambling. The pressure builds, as fear of losing that traction mounts, and heads are put on the chopping block.
4. Missing Browse-only Mode & Feature Flags
Ever been browsing for airline tickets, then go to order and get an error? Try again later? If so you’re familiar with a browse-only mode. This is a very powerful addition to any web application but is very often left out. Some mistakenly believe it won’t work for their application, as users will always be changing data.
Ever visited a website that has star ratings, only to find them missing? Or temporarily unable to edit your rating for a piece of content? This amounts to what’s called a feature flag. These powerful switches give operations teams the ability to disable heavy features, while the side is under tremendous load. They can take a huge burden off the shoulders of your servers when you hit that scalability cliff.
Check this: Why I Don’t Work With Recruiters
5. Operations as an afterthought
I outlined some of the top reasons Why Startups Desperately Need Techops. It is a repeating refrain. Priorities of a growing startup often involve taking on technical debt. But if that isn’t managed carefully you’ll run into some of the problems that Ward Cunningham Warns Us About.