Cookie policy: This site uses cookies (small files stored on your computer) to simplify and improve your experience of this website. Cookies are small text files stored on the device you are using to access this website. For more information on how we use and manage cookies please take a look at our privacy and cookie policies. Some parts of the site may not work properly if you choose not to accept cookies.


Joyent: the application superserver

  • Print
  • Share
  • Comment
  • Save

Jason Hoffman calls himself a ‘data center industrialist’. The 39-year-old co-founder and CTO of the San Francisco-based cloud powerhouse Joyent says his company is industrializing data centers much like mechanization and standardization transformed the shipping industry. Instead of having longshoremen carrying loads to and from ships, we now have heavily mechanized ports designed around the standard shipping container. In the same way, infrastructure cloud for Joyent is taking the job of provisioning a server out of the hands of a human and automating it, shifting the cost from payroll to technology.

Joyent provides Infrastructure-as-a-Service (IaaS) and Platform-as-a-Service (PaaS) via its public cloud built on home-grown hardware and software out of multiple colocation data centers around the world. Recently, the company also started shipping a hardware stack optimized for its cloud platform to service providers that wish to deploy Joyent clouds on premise. It hooks them up with the hardware and its full cloud software package called SmartDataCenter.

The hardware running Joyent’s cloud is based on Intel platforms, and Hoffman says Intel is the only company whose technological innovations Joyent is dependent on (Intel is an investor). It develops everything else in house. “There is no outside software in our software stack,” Hoffman says. Joyent does everything, from kernel to BIOS and virtualization, on its own.


The company is a big supporter of the open-source software movement and as such uses some code developed by the open-source community. Joyent is a major driving force behind Node.js, an open-source software development framework that gives JavaScript developers server-side programming capabilities. The platform is often described as an alternative to the “members-only” Ruby on Rails club.

It is based on Google’s V8 JavaScript engine. Used by the likes of LinkedIn, Yahoo!, Microsoft and Sprint, Node.js supports JavaScript development and management of web applications that serve thousands of users and devices in real time. LinkedIn, for example, reportedly used Node to build its iPad app and Microsoft offers a software development kit for Node on its Windows Azure cloud.

Joyent is the largest contributor to the open-source development project, and employs the framework’s original creator Ryan Dahl. After being project lead for Node since it was launched in 2009, Dahl announced in January 2012 that he would be ceding his “position as gatekeeper” of Node.js to another Joyent employee and member of the Node development community named Isaac Schlueter.

In addition to providing a large pool of capable JavaScript developers Node is attractive in the way it uses underlying infrastructure resources to run software built on top of it. Instead of blocking pre-assigned levels of server resources for a user (independent of actual use), the framework provisions as much server capacity as needed at any given moment.

Joyent’s partner Nodejitsu, a New York City-based PaaS provider offers a platform built entirely on Node.js as a service hosted on Joyent’s public-cloud infrastructure.

Under the hood

This infrastructure supports Joyent’s bread and butter: Its IaaS offering generates the biggest chunk of the company’s revenue.

The infrastructure runs out of retail and wholesale data center provider facilities in the US and Europe. Hoffman expects the company to bring more locations online by the end of 2012. Its current data center providers include Level 3, Equinix, DuPont Fabros and Digital Realty Trust. Total data center capacity  deployed today is in the 10s of megawatts, Hoffman says. A typical single data center deployment for Joyent is 0.5MW, powering about 48 racks that support servers with about 4,000 CPUs on board.

These are “rack-size appliances if one wants to think of them that way,” Hoffman says.

The rack includes standard Intel commodity servers that have been built by a variety of manufacturers, including both original equipment manufacturers (OEMs), such as Dell, and original design manufacturers (ODMs) such as Tawanese Quanta and MiTAC. Each box has two sockets, with 12 cores per socket. Each 2U box will include 256GB of memory, SAS drives and two or four SSDs.

The servers are interconnected with a top-of-rack network switch. A six-rack deployment will require an aggregate switch and 24 racks will call for a core switch. A rack with 16 2U servers will have a power density of about 10kW.

There is no network between compute and storage – a key feature of Joyent’s cloud infrastructure. All storage is local, which is one of the main technological differentiators from competitors like Amazon Web Services, which still has “that same block-storage-over-a-network mentality,” Hoffman says.

“We’re just able to be like the SAN (Storage Area Network) that natively runs VMs (virtual machines) on it. You never have this mysterious network in the middle.” Local storage, he explains, gives the Cloud more reliability (there are no single points of failure) and faster performance.

An I/O bound supercomputer

Hoffman’s original idea when he started Joyent in 2004 was to build an I/O bound supercomputer that would be used to serve applications.

A medical-school graduate, Hoffman spent the majority of his life after college in the field of medical science. A lot of his work involved high-performance computing for medical research – such as analyzing DNA data of cancer patients. This is where he learned about HPC and where the idea for Joyent was born.

“Supercomputers are good at doing math,” Hoffman says. “That’s what they’re for. They’re not good at handling a lot of data.” Supercomputers used in research were CPU bound, and a supercomputer that would serve applications would have to be I/O bound, which means the app-serving computer would have to be designed with an emphasis on I/O rather than on raw processing power.

Hoffman started Joyent with his partner David Young (currently Joyent’s CEO) as a side project. At the time, he was a fellow at the pharmaceutical company Novartis, gave lectures on combinatorial methods at a university and taught people how to get high scores on medical-school admission tests, among other things. “Next thing I knew Joyent was doing really well,” he says. In the end Hoffman chose the job that paid the most.

‘End-to-end’ competition

Today, Joyent’s competitive environment is strong and vibrant. Instead of focusing on one segment of the cloud market, the company has chosen to go to battle on multiple fronts: IaaS, PaaS, cloud software and hardware.

In the IaaS space it is competing with gorillas of cloud-based infrastructure services like Amazon and Rackspace. In selling cloud software to service providers, it has taken on the likes of VMware, RedHat and Citrix.

The company is also facing increasing competition from traditional OEMs in the growing market for pre-packaged hardware-infrastructure solutions for cloud deployment.

Joyent started selling cloud infrastructure solutions to service providers only this year. “You can think of that much like a channel for the service provider business,” Hoffman says.

Telefonica, the Spanish telecommunications company, became one of its first provider customers. (The telco also recently became an investor by participating in Joyent’s latest US$85m financing round in January 2012. Other investors include El Dorado Ventures, Greycroft Partners, Accelero Capital and Intel).

Most major OEMs have either come up with their own end-to-end data center hardware solutions for cloud or teamed up with others to combine technologies into such solutions. Examples range from FlexPod by Cisco, NetApp and VMware to all-Dell and all-HP cloud-infrastructure solutions and beyond.Hoffman’s answer to this is that Joyent can do all of it better and cheaper. This attitude seems to be a necessary attribute for any thirty-something founder of a successful tech start-up, and Joyent’s CTO has plenty of it.

This article first appeared in FOCUS magazine. To register for FOCUS digital editions, click here.


Related images

  • Jason Hoffman, CTO, Joyent

Have your say

Please view our terms and conditions before submitting your comment.

  • Print
  • Share
  • Comment
  • Save


  • Overhead Power Distribution – Best Practice in Modular Design

    Tue, 10 Nov 2015 16:00:00

    Overhead power distribution in your data center offers many attractive possibilities but is not without its challenges. Join UE Corp’s Director of Marketing, Mark Swift, and CPI’s Senior Data Center Consultant, Steve Bornfield, for an exploration of the options and some of the pitfalls, supported by real-life examples from the field.

  • Overcoming the Challenges of High Power Density Deployments

    Wed, 4 Nov 2015 19:00:00

    Increasing rack power densities saves space and energy, and improves both OPEX and CAPEX. But it can also create unintended problems that could bring your data center to a screeching halt. Join Raritan’s VP of Products & Marketing, Henry Hsu, and DCD’s CTO Stephen Worn, as they reveal the three key challenges in deploying a high density cabinet, and explain how to: Reduce operating costs, Increase up-time , Improve mean time to repair, become more energy-efficient manage existing capacity and plan for growth.

  • Squeezing the Lemon - The Power to do More with Less

    Tue, 20 Oct 2015 08:00:00

    Energy costs rising, manpower resources falling – managing a data center is getting more stressful by the day. One cold night could be all it takes to tip your power supply over the edge. And let's not forget the never-ending demands from IT for additional space. More information on its own is not the answer. Join Rittal's webinar to understand how to: • Lower your power consumption and OPEX charges, with 'smart' power distribution • Identify issues before they become problems, with intelligent PDUs' monitoring capabilities • Expand your DC as your business grows, with modular PDUs • Profile your power requirements to help you plan and make better-informed decisions REGISTER NOW Note: All attendees will receive a free copy of the latest White Paper from Rittal.

  • Live Customer Roundtable: Optimizing Capacity (12:00 EST)

    Tue, 8 Sep 2015 16:00:00

    The biggest challenge facing many data centers today? Capacity. How to optimize what you have today. And when you need to expand, how to expand your capacity smarter. Learn from the experts about how Data Center Infrastructure Management (DCIM) and Prefabricated Modular Data Centers are driving best practices in how capacity is managed and optimized: - lower costs - improved efficiencies and performance - better IT services delivered to the business - accurate long-range planning Don;t miss out on our LIVE customer roundtable and your chance to pose questions to expert speakers from Commscope, VIRTUS and University of Montana. These enterprises are putting best practices to work today in the only place that counts – the real world.

  • Power Optimization – Can Your Business Survive an Unplanned Outage? (APAC)

    Wed, 26 Aug 2015 05:00:00

    Most outages are accidental; by adopting an intelligent power chain, you can help mitigate them and reduce your mean-time to repair. Join Anixter and DatacenterDynamics for a webinar on the five best practices and measurement techniques to help you obtain the performance data you need to optimize your power chain. Register today!

More link