ISP Failures and Ususpecting Users?

cybermage asks: "With the cash crunch in the .com area, I'm wondering what will happen with companies like this. These guys are a local ISP that borrowed a lot of money to roll-up smaller ISPs and now has a mountain of debt and maybe not enough income to justify the debt. What's to become of their 15,000+ dial-up customers when the money stops. Given the lack of regulation in the industry will there be a lot of sudden ISP failures and what do people think the backlash will be?"

The Death of the smaller ISP...

[crisco]

...has been predicted for years. It hasn't happened.

In the case you mention, it sounds like bad business practices. Sure, they might go under and a bunch of people are left looking for something else, but thats just an opportunity for some other savvy techs to snag a good chunk of them and keep them happy. Sure, they'll lose some to the big guys, but the small and medium sized shops will stick around.

respectible thing

[whydna]

The respectible thing would be to make arrangements with some other (perhaps larger) ISP (such a Mindspring, etc.). Ideally, the users' accounts would be transfered to the larger ISP. Perhaps some of the equipment might go too.

I suppose a larger ISP might be able to provide some sort of re-imbersement package for the user accounts/equipment... maybe not... I dunno..

just a thought

Re:Sell assets to another ISP

[Baz Quux]

(if the username at the new ISP was taken, we had to call the customer and have them choose a knew one. Nobody was very happy about that).

Yikes, I don't suppose they would be.. We rarely had cases where that was necessary. We made hacks for both the UNIX servers and the billing database to avoid ever having to do that for more than a few select cases. This involved clusters of Navis Radius servers using realm and DNIS info to associate a user with a particular ISP, and auth off that ISP's shadow/users file, and tacking a dot and 3-character extension to the usernames when merging them into the main billing database. For email, we just left each original mailserver running, but were working on a modified q-popper that tied into the Radius system to work similarly to the dial-up authentication, and we would have been able to merge whole mailservers into the system.

Man, having to contact all of the static-IP users was bad enough, I feel for you having to call anybody whose username collided!
--

Doesn't sound like many customers to me...

[Baz Quux]

So, they borrowed a bunch of money, set up a 6,000-sq.ft. co-lo datacenter, built their own fault-tolerant, fiber-optic backbone, and ended up with only 15,000 customers? (Where did that figure from above come from anyway? I couldn't find it on the link given.) That's awfully low.

You could easily put 15,000 users on a few Ultra-II's, a couple of DS-3's, and maybe a couple of Ascend MAX-TNT's, and not need the overhead of giant co-lo facilities and expensive networks. That's less than $1M of gear, plus monthly operating expenses, which I would expect to be sustainable.

I spent the last year working as a sysadmin for a company that operated similarly in the southeast, and had some comparable - if higher - dollar figures, but we had 50 ISP's and over 300,000 customers. When I left, they weren't in great shape, but their head was still above water. That many customers can provide an awesome amount of cash flow. Being tasked with sending out the global email to all customers, announcing our impending doom, is something I never had the joy of experiencing (ha!).

In the end though, it's a question of whether or not the company that exists is worth: a.) operating as it is, b.) absorbing it into yet a larger provider, or c.) simply letting it disappear. In the first case, some management change-outs and heavy infusions of capital could bring the company out of the red long enough to make things self-sustaining once again. Some employees end up on the street, others make it big, but the customers shouldn't notice much, if anything.

If their network and most of their kit is crap, though, some larger company may or may not decide to risk that aquisition. Depending on their skill, such a transition could go smoothly and unnoticed, or horribly, and involve lengthy outages. It could be as simple as putting LMP's on all the dial-up numbers to point them at your own gear, and migrating webservers and mailservers, or it could be just a total unscalable mess.

If so, they may close the doors and go away. I would expect customers to be notified at that point - there is no face left to save by then. The assets (servers, routers, workstation, miles of Cat-5) will get liquidated, circuits turned down, power shut off, and the locks changed before the old datacenter gets turned into a Gap or something.
--

Re:Doesn't sound like many customers to me...

[cybermage]

(Where did that figure from above come from anyway? I couldn't find it on the link given.) That's awfully low.

The # is based on speaking with some of their past employees. They mention 1,000 co-located servers, but that's not right, either. They simply spent more on acquisitions then they should and have more debt then they can hope to cover in revenue, (running at seriously negative EBITDA).

--

Re:Sell assets to another ISP

[cybermage]

It is quite possible that the value of the customers doesn't equal their debt. Since they are a public company, they may not be able to simply sell off assets to cancel debt and end up still in operation. Their best bet may be to file for bankruptcy liquidation, but this would stand a good chance of simply interrupting the service for their customers while the liquidation gets handled. Selling off the assets without bankruptcy will probably get them sued. Kind of a catch 22. The customers are worth more, if their still operating, but may be unsellable unless the whole company is bought (not worth it.)

--

Re:The Death of the smaller ISP...

[shippo]

A few ISPs in the UK have gone under. One notable one provide unlimited free call access for a single fee - usually in the UK local telephone calls are charged at a per-minute rate.

They had serious problems from the outset - staff were not being paid on time, treated like dirt, that sort of thing.

Eventually they got a bill from their provider for around 2 million GBP, and realised that they couldn't pay it.

Sell assets to another ISP

[Hyperbolix]

I used to work for a small ISP that had been forked off of a profitable consulting firm. We went under because the ISP part had only about 1000 users, and equiptment costs were just being pulled from profits of the consulting firm. The balance of profit/spending died, and so we sold our customer base to another larger local ISP (about 10000 users) which was basically the same setup (ISP branch from consulting firm). I was offered a job there, and the customer base was migrated in a very, very gruelling process (if the username at the new ISP was taken, we had to call the customer and have them choose a knew one. Nobody was very happy about that). As for equiptment, our modem banks were not used at the new ISP because they took ISDN wiring and it would be to expensive to rewire to accomadate for that. Some big technical reason that was hardware related, which, I, as a poor sysadmin, couldn't really understand. Yup. It actually happens pretty often. At the new ISP, we gobbled up a few others also while I was there.