Redundant Credit Card Processing Solution?

RokaMoka asks: "As I type this, I'm on hold with Verisign Payment Services, our (only) merchant services provider. I run several e-commerce sites, and how shall I say... 'tis the season. At the moment, VPS is totally down, and I am losing thousands of dollars per hour. Does anyone have any experience in designing and supporting e-commerce solutions with multiple vendors for CC processing? What other networks are out there, and what has been the customer experience with them? What should the strategy be, load-balance or fail-over?"

Queue the Transactions

[Logreybaby]

Rather than failing to authorize the CC when VPS is down store all the needed information in a table, queue, etc. This allows the user's transaction to complete and it allows you to authorize CCs in a batch process once VPS is back up. Obviously this is not how you should run all the time, just when VPS is down.
BTW: I would not ship anything until I successfully authorized and charged the CC.

Load balance or failover?

[FooAtWFU]

Well... I'd think it's simple. Which is cheaper for you to run, load balance of failover?

Eh?

[Cthefuture]

"Thousands of dollars per hour" and you can't afford a direct merchant account with the various CC companies?

They will guarantee a much higher level of service than going through some 3rd party.

If you need to, hire someone to hook your mechant account to your web sites. Simple as that, you got the money.

Get your own system

[hectorh]

Any business that is running several e-commerce sites and processing thousands of dollars per hour should be using their own credit card clearing system.

This is when people realize that the lowest bidder is not always the best choice.

Load Balacing

[New Breeze]

You're going to get killed in fees if you don't process a decent level of transactions through a backup processor when it does come into play.

As I type this I have a client who's CC processing has been down nearly 24 hours, and has resorted to a dial backup solution. Not exactly the way to process 5000+ orders a day. And to top it off they sent out a special email offer to 500,000 subscribers this morning, so they're dying as we speak, and if it's not resolved in the morning we may be switching providers in a hurry. Thank the stars that they choose their own provider...

Ignore the posts talking about why you don't need this, and SLA's. No SLA is going to replace lost revenues, and anyone who doesn't have a backup plan in place is just waiting to get burned.

a few comments

[gradbert]

Firstly GGIAITCCI (golly Gee I AM in the credit card industry)

To those posters who thing that "thousands of dollars per hour" is large enough to justify processing credit cards yourself, that is really not a large number. $1000 /hr is only 8.7mil a year. Your local MegaLoMart probably does more than that. They are probably paying around 100,000 for processing. Doing your own processing would require an investment over a million dollars and sponsorship from a bank.

Having a second credit card processor would be a mild pain. it would probably require having two merchant accounts with two different banks (since a processor can only process merchants that are with banks they have arangments with). Not to mention that you would have support two different interfaces.

Your best bet would be to switch to a processor who takes downtime seriously.

Change processors

[Ender_Stonebender]

Disclaimer: I work for Fifth Third Bank, in a credit card processing capacity (software testing and installation for authorization/settlement system).

First off, Verisign being totally down is completely unacceptable. Demand a refund for the service outage.

Second, why the hell are they totally down? The system that I work with (one of several owned by Fifth Third) is never completely down. We have three access methods; dial, SSL, and non-SSL TCP/IP. It's rare for one of them to have problems, virtually impossible for all of them to get hosed at once. We run on Tandems, which allow for "buddy" process running in seperate CPUs where the secondary takes over if the primary has a hardware problem; we have redundant access to our disk drives so that we can always get to the data. We also have a voice-menu system that you can use to authorize (not a good plan for e-commerce, but I figured if I was plugging the company I work for, why not?). Hell, we even have two identical systems in widely seperated locations! If you can't get through to us, you've probably got bigger things to worry about because there's been a major natural disaster.

Third, WTF did they change during the holiday season that blew up their system? We have a concept called "peak season freeze". Basically, we change *NO* software or hardware between mid-November and the end of September, except emergency fixes for things that are totally broken, and even that is rare.

Fourth, the guy who said you should running your own credit card processing solution is an idiot. He obviously does not know how the credit card processing world works and has never attempted a certification with one of the credit networks.

--Ender
PS I'll go write up an explanation of how the credit card processing world works in my journal now, so that you can go educate yourselves on the basics.

Simple CC Vaidation

[Unholy_Kingfish]

This time of year isn't our busy time, but we have it other times of year.

We rely on monitoring the sites, and if there is a problem, switch from the processor to a simple CC capture. We would have to process the orders by hand, but we would only lose the sales that occur between the event of failure, and the switchover.

The key is knowing of a failure, and switching over.

As far as having two gateways/processors. This will be tough. You could have two of each, and just switch to the other one if the other goes down. Your store software should allow you to change this easily. Just disable one, activate the other. Problem with that is having a second gateway and processor which you don't use, then all of a sudden use, the will charge you out the ass.

Now you could get two gateways, and ONE processor. But if the processor goes down, you are completly screwed. But in my years of dealing with gateways and processors, it is 999/1000 the gateway's fault, not the processor.