JetBlue TrueBlue - Hey JetBlue - You need a bigger server

Go try to check the status of a flight today (there is a snowstorm in the North East).

You can't! The server is too busy.

Jetblue needs to do what Ticketmaster did. Base your server needs on worst case (volume) scenario. It's days like today that you need to allow your customers to do as much as possible online and not be forced to call your 800number.

Yes, smaller airlines like B6 and F9 have this problem when a situation like today's snowstorm just wipes out their computers. Having enough capacity would really help them get through these types of situations without ticking off as many of their customers.

OT, the Colorado Rockies learned this the hard way with World Series tickets as well. I do not buy the "malicious attack" story, it was just a whole lot of people trying to get tickets that crashed the server.

Did you find the information you where looking for? My flight/trip was cancelled so I'm sitting around on reserve with nothing to do

FYI most evening operations out of BOS have been cancelled and JFK operations are running on a delay with some cancellations

FYI most evening operations out of BOS have been cancelled and JFK operations are running on a delay with some cancellations

Yes, because B6 has been shutdown at JFK & BOS due to snowstorm. I think all flights will have been canceled today. It should be bring it back normal operations tomorrow or Sunday. If you wants to do something to refundable your money back. It's guarantee B6 will have your money back for your future flights.

It was really bad here in BOS. That and it being Presidents Day Weekend we had loads near capacity. My trip was cancelled so I'm sitting standby in hopes of going some where or being released to go home. It looks like I may have been assigned a SJU redeye turn tomorrow night but it looks like we'll be back to normal tomorrow if the weather moves out like it is suppose to.

Yes, smaller airlines like B6 and F9 have this problem when a situation like today's snowstorm just wipes out their computers.

If anything, it should be easier for a smaller airline to NOT have these problems since its easier to scale up to the processing power of a larger airline even though B6 doesn't have the customer count of a larger airline.

It's not a single server, it's 12 servers at the moment.

And actually, there are other factors that go into the loading. A single query once brought down most of the entire system as it tried to load every single passenger into memory at once.

If anything, it should be easier for a smaller airline to NOT have these problems since its easier to scale up to the processing power of a larger airline even though B6 doesn't have the customer count of a larger airline.

I wholly disagree with this view of things. A larger company has more infrastructure in place initially and is more likely to have the ability to shift resources around to address specific load demands as necessary. A smaller company has more finite resources and will have more difficulty trying to scale up. For a larger carrier the incremental volume they need to scale by probably isn't as great as for a smaller carrier, so the budget to be able to scale isn't as significant.

My favorite example of a well designed web server farm for sccalability is espn.com. Like most other major web presences they have a farm of servers to handle the requests coming in, but the architecture is such that specific servers handle the different sports categories. This allows them to add more servers to a specific area as needed, rather than throwing generic processing power at the whole site. Right around now a number of servers will be devoted to NCAA basketball (March Madness), and then a few months from now they'll be moved to NHL/NBA playoffs and then to MLB for the summer run and then to NCAA football and the NFL later in the year as those sports ramp up. It is a pretty impressive architecture that they use to meet the needs of their users.

And actually, there are other factors that go into the loading. A single query once brought down most of the entire system as it tried to load every single passenger into memory at once.

That sounds like a programming issue to me (ie allowing such a query to be run against your production servers/dataset).

I wholly disagree with this view of things. A larger company has more infrastructure in place initially and is more likely to have the ability to shift resources around to address specific load demands as necessary. A smaller company has more finite resources and will have more difficulty trying to scale up. For a larger carrier the incremental volume they need to scale by probably isn't as great as for a smaller carrier, so the budget to be able to scale isn't as significant.

Unlike your ESPN example where demand can be somewhat tough to pin down, the number of requests to any airline is pretty much relational to the number of flights and passengers.

In other words, if B6 has 1/2 the flights/customers of a larger airline, its fairly easy for them to have at least 1/2 the processing power of that larger airline. As the smaller entity in the equation, it is definitely easier for B6 to scale up. There isn't a logical reason for not having the processing power required when the number of queries is fairly easy to calculate; its not like when a snow storm hits everyone is going to all of a sudden think of buying tickets on B6 and clog the system.

Assuming the new beta flight notification feature works properly, it will take some of the load off their system.

In other words, if B6 has 1/2 the flights/customers of a larger airline, its fairly easy for them to have at least 1/2 the processing power of that larger airline. As the smaller entity in the equation, it is definitely easier for B6 to scale up. There isn't a logical reason for not having the processing power required when the number of queries is fairly easy to calculate; its not like when a snow storm hits everyone is going to all of a sudden think of buying tickets on B6 and clog the system.


I'm not sure I follow this. It is very easy to predict within a reasonable margin of error the normal load and plan for that. In this case the number of queries in the case of a weather problem is harder to predict. Could JetBlue just predict that every passenger is going to run a query at the same time and spec accordingly? Will UA have the same percentage increase in load? Same absolute increase? It isn't a cut and dry answer, IMO.

Scaling up is theoretically easy and just requires more servers (more or less), but having extra servers running if they aren't needed costs real money. It is a balance and B6 lost this time around. Maybe they'll grow the server farm, maybe not, but they know what the issue is and how to address it.

as is publicly available knowledge (I'm not revealing anything here that isn't in the public domain), Jet Blue doesn't run their own reservations system. They outsource to a company that does res systems for many airlines.

On the scalability issue, this just isn't the case. You can plan to a certain degree but it really boils down to which queries are running against which flights and how many people are actively using the system. You can have X number of queries run on 2 different days and they will take wildly different amounts of time to run depending on what the query has to reference. Jet Blue uses a server farm (again publicly available information via microsoft's case study) that can scale up to as many servers as they wish to put in.

As for the developer/consultant that wrote the query that initially loaded every customer into memory at once. He no longer works for the company. Some of you may or may not remember back a year and a half or so ago when JB had massive computer outages for a day...

As for scalability between the larger airlines and JB, again this just isn't the case. It comes down to implementation. I worked/programmed for Sabre a while back and they had huge mainframes that processed a lot of stuff. Adding capacity to those systems was EXTREMELY difficult. Since JB uses a server farm with very scalable software, they could theoretically add more processing power than even the largest airlines in the world.

Since JB uses a server farm with very scalable software, they could theoretically add more processing power than even the largest airlines in the world.

Really? I would have never guessed that. ;)

Seriously, thanks for adding some additional perspective. ^

Here's one article: http://findarticles.com/p/articles/mi_m0EIN/is_2003_April_24/ai_100563122

a bit more recent information.

http://www.microsoft.com/casestudies/casestudy.aspx?casestudyid=200076