For how many of you out there is running a firewall—or running IT in general—central to your business? I bet the percentage is a small one. Nevertheless, you have to do these things in order to protect and facilitate the things you really do for a living.
This is why I believe in services for most IT functions. By “services,” I mean that a function should be outsourced to some outside firm. Web hosting is a good and obvious example of a service. You could run your Web site on your own systems, but wheres the sense in that? Hosting firms have great economies of scale when it comes to running large numbers of customer sites on a single box as well as managing farms of Web and other application servers. As a result, its cheaper and better to outsource it.
In the case of security, I think there are also economies of scale, at least in many cases. Years ago (I think it was 1999), I wrote a story predicting that ISP-managed services were the future of antivirus and some firewall functions, at least for consumers. If spam had been so big a problem back then, I would have included it, too. For the most part my predictions have been flops, but I still think I had good arguments:
- Economies of scale: An outsourcing firm can throw a few large servers at a problem and support large numbers of users.
- Timeliness: One of the key characteristics of good antivirus, firewall and spam-protection support is having and implementing the latest attack information. Its a lot easier for a service to update its server network than for a company like Symantec to push out updates to millions of users.
- Performance: In some cases, especially spam filtering, outsourcing means youll cut the amount of traffic entering your network, leaving only the parts you want. At the very least, this move will improve bandwidth. I wonder how much hardware and bandwidth money could be saved by using managed security services.
- Backup: Service providers make guarantees of backup, archive and disaster recovery.
I think those factors are all the biggies, and they are even more true now. So why was I wrong? I think the major reason why the service model never caught on, especially when it comes to ISP-provided services, is that people are cheap. Yes, you heard it here: ISPs assumed that customers wouldnt be willing to spend another $5 per month or whatever it would take, so they never rolled out the services.
Ive also always been suspicious of the motivations of antivirus companies on this front. If ISPs generally adopted antivirus scanning, it would certainly hurt the retail antivirus market. That money would be somewhat offset by the subscription fees, and the service model is also far cheaper for the antivirus vendor than shipping millions of boxes to consumers, but it has to be a scary move for a McAfee or Symantec. On the other hand, for a company like Sophos, which is respected in the corporate market but has no meaningful consumer business, this would be a pure win. Any antivirus vendor would point out that a service that scans mail, maybe even HTTP, still doesnt stop all possible avenues of infection; therefore, you still need to run a local scanner. But lets face it: E-mail is where real people get their viruses these days, so plenty of people would conclude that they didnt need to buy or update a local scanner anymore.
Continued on Next Page
More on Security Services
I mentioned performance as an advantage, but of course, theres also a potential performance downside to this approach. Outside systems introduce a latency that is somewhat outside your control. With respect to functions such as e-mail, I think people understand that a certain amount of latency is built into the model, and they accept it. However, its reasonable to expect a certain level of performance. Introducing a 15-minute delay in e-mail would be bad, for example. Such service-level agreements (SLAs) typically also include guarantees of uptime to a certain level.
My own ISP, Speakeasy.net (a DSL ISP using Covad circuits), is something of a pioneer in this area. For some time the company has offered a semi-managed firewall and it is on the verge of releasing anti-spam and antivirus support. But antivirus support at the ISP level (which usually means e-mail antivirus) has been rare among ISPs, although Yahoo! Mail and Microsofts Hotmail have had antivirus scanning built-in for years.
Now things are somewhat different when it comes to spam. A growing number of ISPs claim to provide spam-blocking capabilities. There are also numerous corporate spam-filtering service solutions, such as FrontBridge Technologies. FrontBridges solution does everything you would want to do and many things you probably dont have the resources to do; for example, its antivirus checking for e-mail goes through AV engines from multiple vendors, and it checks for updates every 10 minutes. You control the criteria for evaluating spam, including your own whitelist, proprietary blacklists and a rules database. FrontBridge has seven geographically distributed data centers on multiple backbones for better redundancy and performance. Finally, if your mail server is down, the company will cache and queue your mail for up to five days.
Is it cheaper? Thats hard for me to say; FrontBridge prices based on bandwidth (about 30 cents per megabyte) or by user (about $2 to $3 per month per user). I can easily see that being cheaper, especially if you can treat the whole service as an expense as opposed to having to depreciate your own equipment.
Services have another advantage: Providing for many customers gives them a perspective that can help everyone. I-TRAP Internet Security Servicesoffers a firewall monitoring/intrusion-detection system that uses a customized, hardened Linux server on the customer premises, but the real smarts are on a server back at I-TRAP. Because it sees so large a set of attacks and traffic it can apply statistical analysis to events at your site to gauge their seriousness. I-TRAP also provides sophisticated reporting capabilities that are easy to update because they are on a central server.
Serice providers usually claim that their offerings save their customers money, and perhaps they do; I havent run the numbers, so I dont know for sure. But even if they cost the same, there would still be advantages to the service approach, because it simplifies your own business by not having to manage these things yourself and not having your own servers directly exposed to the Internet. It also affords you flexibility. Its probably easier to switch between service providers than it is to change security software youre running in-house.
I still think that one day consumers will buy into these security services and everyone will be better off for it. That day doesnt seem a whole lot closer today, but it still seems inevitable.
Security Supersite Editor Larry Seltzer has worked in and written about the computer industry since 1983.