Open source faces hurdles in the cloud

It may not be efficient to open-source some cloud functions, and large vendors will 'dominate' cloud computing in 2020, according to Mimecast's Nathaniel Borenstein
Written by Jack Clark, Contributor

As companies gain ground in cloud computing, a debate is growing over the extent to which cloud companies should collaborate on open standards and open source to protect customers from vendor lock-in.

In July, a consortium of cloud companies launched the OpenStack scheme to spur the debate and give the cloud industry extra impetus. Companies are discussing the value of opening up: some are worried it could hurt their competitive advantage, while others see it as an opportunity.

ZDNet UK recently spoke to Nathaniel Borenstein, one of the co-creators of the integral internet communications protocol Mime and chief scientist at Mimecast, a cloud-based email management and security services provider. One of Borenstein's main jobs at Mimecast is looking at trends, patterns and movements to work out what the cloudspace could look like in 2020.

Q: At Mimecast, you're working on predicting the space — and the opportunities — for 2020. What do you see for the cloud?
A: I think that the market imperative that causes large vendors to dominate most tech areas is not going away for the cloud. In fact, it might even be stronger, because a cloud provider in the UK could serve the entire world.

I do expect it to be dominated by a number of large vendors, but I do think some of the large vendors will dominate very specific niches. I think that come 2020 there will be not that many cloud vendors, but I don't expect our competitors to be only doing what they're doing now.

I think there are some adjacent areas that, very naturally, connect. For example, what is a logical closure, in the mathematical sense, of what Mimecast does? No specific products, but things like doing for other forms of communication what we currently do for email. Or adding email services other than the ones we currently have.

Is it right to go for open standards across the cloud industry, as OpenStack has recently committed to?
I think it depends on the application, I really do. They're [OpenStack] providing a virtualised environment in the cloud for running applications more under your control — standards for that, I can see how they can work and they make sense.

But in our [Mimecast's] case, I can imagine some standards that might help a little bit, but the fact is email is pretty well established. In the Microsoft world, it's a PST file; in Linux world, it's a TAR file.

And what about open source within the cloud industry?
The open-source movement started in the 1980s after computing had had a couple of decades to figure out "this is an OS, this is a mail tool". We know now that cloud computing requires a lot of different programming skills. We've found a lot of the open-source would not fit in the cloud, because it wasn't efficient enough.

I was surprised at this when I came to Mimecast... I'm a big believer in open source, but I don't believe it's a panacea. I don't believe it cures everything. I try to avoid extremes here.

I do think that in the cloud world it is likely that a lot of vendors will have a lot of their intellectual property wrapped up in one place in the world — their own servers. That may not be that useful to other people. And [for the company], that's like open-sourcing your wallet.

Rackspace's head of cloud Fabio Torlini has said that open-sourcing some of Rackspace's core intellectual property — file servers and file storage — was "a win/win situation for our competitors". However, the company went ahead and did it because it believes it will have a competitive advantage when it comes to the development of the software, even if it's open source. What do you say to that?
I'm sure there are some things that this is true for, but to the extent that our system is all devoted to a single major purpose — email — it may be hard to extract bits of general utility. But if we open-sourced the whole thing...

...we'd end up fielding support queries from our competitors. It's not a pleasant situation.

Most parts are very specific to our application — we're email specialists. If we open-sourced this system, then it would be difficult, because it's a smaller sector and these are our competitors. But at Mimecast, we have been talking about open-sourcing a couple of tools.

Can you tell us any details about the items you are thinking of open-sourcing?
It's more specific to email than cloud. Its connector level stuff that reaches out to other databases and does things. I believe it generally runs locally, but what it's good at is decoding difficult formats for use in other contexts. It's not a cloud function — it's an auxiliary piece of code that we use for supporting our cloud offering. We'd like to open-source something, because we are believers.

What do you make of schemes like OpenStack in general? Do you think they have come at the right time? How could OpenStack deal with some of the other major companies in the space — Oracle, for example — if they do not come onboard?
It's certainly possible it could be successful, but it's not clear to me who the target audience are. To me, the target audience seems to be much better suited to Amazon right now. If you're a small service provider, it's a better bet today.

After five years, when Rackspace produces this stuff, will it be a better choice? I don't know.

Open source is much more successful in some types of software than others. I would point to the difference between success on the open-source server side with Linux, and then on the desktop side with crap [where there are] few things offered. Out of all these server things, the biggest cost will be running the system, not building it. So making the software free... helps.

The answer might be different for different kinds of companies. One which is very big but not technical — for example, insurance — might have the wherewithal to pick up the open-source thing along with it. A law firm — no way. They can't even run a mail system, that's why they flock to Mimecast. The fact is, at a law firm, you have a few guys running the email system and not keeping up with the threats, and [running] a cloud is several magnitudes harder.

How has the development of Mime informed your thinking?
I think both that work and the work I am doing today is the same basic drive. To some extent, the drive has been to allow people to work together at distance.

Email is a critical technology for doing that. The impetus for Mime came from two communities: the non-ASCII community, who were interested in other languages, and the one I was part of: the multimedia community. Moving those two together picked up the whole thing, this thing called the world wide web.

One of the things that was driving me was someday I wanted to have grandchildren and get pictures of them by email, and in 2009, I did.

Beyond that, the desire to make remote collaboration possible continues to underlie just about everything I've done. Mimecast has the ability to build a set of services on top of the engine it has, and the purpose of those [services] is to enrich collaboration.

The best advice I received was from a fellow named Dave Crocker. When we were working on what became Mime, he took me aside one day and he said: "Nathaniel, you ought to come up with a clever acronym". He pointed out to me things that did and didn't have clever acronyms, and so I came up with Mime in 10 minutes.

Editorial standards