It’s been a year since we launched blitz.io, an awesome multi-tenant application performance testing/monitoring platform running on AWS and Heroku. Looking back at the year, it’s been an amazing ride and we’ve helped a pretty diverse class of users that have no intent on becoming performance experts to really understand the difference between concurrency and hits. But I’m pretty disillusioned right now. Not to mention bored. And I think its the cloud.
Cloud and the Closet
blitz.io has no IT, no racks of servers, cables, heat sensors, rain-proof roofing, circuit breakers and a whole class of other paraphernalia that are good for job creation. Instead we use the dreaded cloud. Running dozens of servers in all eight regions of AWS, blitz.io is everywhere, yet nowhere. This is not right. Look at the unemployment rates everywhere. The cloud is not helping. While we are able to do more with perl 1-liners that make @devops_borat happy, this can’t be the future.
IT as an entire discipline is dying. Let me ask you this. Can you put a mug of hot piping coffee on 2U rack mount while you spend the morning tracing which port of which switch has this said 2U connected? What about an ARP flood that takes out the entire local LAN? The cloud sucks. I don’t remember AWS going down because of an ARP flood. EBS maybe, but not ARP. Just a few months ago, the entire Mu engineering team was tracking down an ARP flood by yelling and shouting and just bonding as a team. Now we just stare at our perl 1-liners and yearn for those floods.
Auto scaling sucks
Whoever invented auto scaling should be shot. Looking at you Jeff Bar. Whatever happened to pleading IT to upgrade the CPU and memory on a production server, going through Outlook scheduling fun, a run to Fry’s, announcing a blog about scheduled downtime and all of that? Now a couple of rules later, up pops a bunch of EC2 instances and an hour later away it goes. Bunch of virtual particles. If you weren’t looking at the AWS console (or your bill), how do you know auto scaling is working? I say, this is the worst EC2 feature ever. It’s completely against ITIL, change management and what a way to eliminate a glorious discipline called IT?
Our web tier runs on Heroku and while you think cloud sucks, PaaS makes me puke. And here’s Adam Wiggins going on and on about his 12 factor BS, software attrition and more BS. I’d say running out of disk space on a server because you forgot to setup logrotate is character building for devops! Heroku apparently thinks logs are streams. What a stupid idea! And all of those add-ons that take configuring sendmail and setting up postgres servers and tweaking configuration parameters a moot point? To hell with that. This is the worst a company can do to encourage unemployment. Bad for the economy, I tell you, bad.
You know, when I founded Mu, we used Perforce and then so we don’t have to pay for licenses, we switched to Subversion. And along came GitHub making coding collaborative. What a bunch of ballooney. Anyone that hasn’t experienced checking in code over a network VPN to the corporate doesn’t deserve to be a programmer. What’s up with integrated issue tracking, pull requests and all of that? My best part of the year was when GitHub got hacked with the Rails vulnerability. Should teach them a lesson in having a cloud-based product. If I were them, I would kill that and just support the enterprise on-prem version instead. And really, who called for an API on a revision control system? Seriously?
I’ve got nothing more to say other than wishing Damien would rewrite CouchDB in Objective-C. Enough said. And I think we need to have more votes before releases. 5 is too little.
Well, there’s more to rant, but I think all these cloud-anything-as-a-service suck. And it’s bad for startups, not to mention the economy. And any founder that wants to build a company that lasts, would run like hell away from these solutions. And yeah, happy April 1st.
If you got this far, you might also like my blog on ‘Why NoSQL is bad for startups‘.