Heroku adds Big Gulp 'XL' option to cloud menu
Salesforce-backed cloud boosts capabilities, but keeps prices high
Salesforce has upgraded its Heroku cloud to take advantage of more powerful technology in the Amazon Web Services infrastructure upon which it floats, as it tries to support larger, more complicated customers.
Heroku is an Amazon-based application runtime service that takes pride in its ease-of-use for developers who prefer to focus on applications rather than infrastructure. It supports multiple high-level languages such as Ruby, Python, Node.js, and Java.
Heroku's technology lets people get applications up onto the service quickly, but also lets devs who have no familiarity with infrastructure perform the cloud equivalent of running with scissors – a fact that became apparent last year when Heroku admitted that its customers had been subjected to degraded performance for the prior three years due to a bug in the service's routing framework.
The company announced its new"XL" edition of the Heroku platform-as-a-service on Monday, and seeks to fix some of these perceived issues by adding in greater diagnostic information, a 24/7 support option, and more powerful rentable application servers.
These upgrades are billed as being built for applications that routinely field thousands of requests per second, such as the internet's premier mind-bleach purveyor of the moment, Upworthy.com.
"XL" sees Heroku field a new plus-sized unit of software and compute named a "PX Dyno", which has 6GB of RAM, costs $0.80 per hour, provides single tenant access, and is based on the
AWS c1.xlarge instance.
The new service also adds more detailed resource utilization figures for the Postgres database, including metrics such as Index hit rate, Cache hit rate, Database size, Load, Memory usage, and I/O Operations.
It also adds in some pricy paid support options, such as a Premium Support option for $1,000 per month or 20 per cent of account spend (whichever is greater) that guarantees 24/7 support and a 1-hour SLA for critical tickets. A Technical Account Management option runs $1,000 a month in addition to the Premium Support cost, and adds technical consultation with a dedicated support engineer.
As Heroku is based on infrastructure rented from Amazon Web Services, it's obvious that all these services represent a major markup on the base AWS prices – the
c1.xlarge instance, for example, can cost as little as $0.3529 per hour when bought under a three-year reserved instance contract.
It's also concerning that although Amazon has dropped prices numerous times over the past few years for compute and storage, Heroku's prices have barely shifted – leading us to wonder whether Amazon's price cuts are juicing Heroku's profits.
"It's understandable folks are going to look at our prices and then Amazon instance price," Heroku veep Adam Gross, explains to El Reg. "What we need to do a good job communicating is everything else you're getting as part of the Dyno cost – runtime metrics, things like the Websockets we announced a few months ago. The Heroku platform itself has added a lot of functionality over the last three or four years. [The] only place that functionality is reflected is in the Dyno price."
Gross said Heroku is aware of some mutterings of discontent about its pricing approach, and is going to look into this.
"I look at the cloud app and infrastructure market as a whole, and I think 2014 is going to be a really important transition year," he says. "The model of building and running your own data center – I think that model has never been more discounted and discredited than it is right now. All of that energy is moving into the cloud market."
Though the cloud has a large amount of benefits – contract flexibility, good access to new technology, downward pricing pressure – we're yet to see any truly convincing evidence that a cloud-based infrastructure is cheaper for a known, regular workload than on-premise kit managed by a knowledgeable sysadmin. With Heroku XL, Salesforce is betting that it can persuade companies otherwise. ®
Sponsored: Hyper-scale data management