Globex Down Again
- Thread starter SethArb
- Start date
I agree with you. The Globex uptime track record is horrendous. Ask anyone who deals with enterprise class systems with high uptime requirements.Quote from SethArb:
not the first time ...
and probably not the last time either
hopefully the CME will use some of the proceeds
from their IPO this month to do what it takes
to make the Globex System 99 % reliable
if there is such a thing as 99 % in real life
I do not know
And FYI, 99% uptime sucks! That's equivalent to 5,256 minutes of downtime per year. Industry standard best practices is 99.96% uptime which is only 210 minutes of downtime per year. You can get this with a redundant everything architecture and enterprise clustering software. And we're just talking standard enterprise applications here. Order of magnitude software costs in the low six figures.
If you're talking a stock exchange you need to push that envelope to past "four nines" or >99.99% uptime. The cost of course increases exponentially as you get closer to 100%, but it's not like they are hurting for revenue. The cost of downtime for them has to be much greater than cost of maintaining a highly available environment.
Of course, since its inverse would be logarithmic.
Hmmmm, let's see. If they are down in the evenings like tonight, they are probably missing out on 100 contracts traded, so if they are normally down twice a month, an investment of $50,000 to fix the problem would pay off after only 20 years! I can't imagine a reason in the world why they don't go for it...
The problem is you don't know when downtime will occur. 30 mins of downtime may not hurt them at all in the evening with low volume, but what if that occurred at the height of the trading day?
It's a simple business decision. You look at your estimated cost of downtime (not only from lost revenue, but also from marketing/PR perspective, and liability -- the lost revenue is the easiest # to calculate and is always your base value for initial calculations). Then you do your standard consulting matrix and figure where your sweet spot is between 99.9x and 99.99y. The costs get more prohibitively expensive as you get closer to 100%, but there's a sweet spot for every business service. The CME just has to determine what's right for them.
Quote from SethArb:
not the first time ...
and probably not the last time either
hopefully the CME will use some of the proceeds
from their IPO this month to do what it takes
to make the Globex System 99 % reliable
if there is such a thing as 99 % in real life
I do not know
Good luck - the IPO money flowed into people's pockets, not infrastructure budgets. Globex has problems because the head of electronic trading systems at CME is the poster child for the Peter Principle and is flying by the seat of their pants.
Maybe now that the IPO's over, some of the top brass will leave and the CME can actually hire some people who know what they're doing - hopefully starting with cleaning house in departments like electronic trading.
I hope so too. In all of their prospectus' and other information relating to the IPO, one of their #1 priorities was improving reliability. I'd be really curious to see how what their plans are on the technology front. Afterall, an exchange is all about accuracy, reliability (i.e. uptime), and speed (others may have different priority order). This is all magnified when you have an automated electronic exchange.
They should be hiring the best high availability/fault tolerant experts they can find. I know some, they should call me.
