Major Geocaching.com Site Enhancement

[+Captain Morgan]

What I suggest, would cost a lot of dollars, but would be the only solution in a long run, to ensure 24/7/365 services from geocaching.com

 

Instead of few servers in Seattle, USA, geocaching.com should have 1 or 2 mirror sites around the world. Generally, a mirror site holds copies of resources that originate on remote, and possibly far away, sites. Individuals can use a mirror service to access these resources locally thereby reducing network traffic and requests to the originating servers. As a result, users enjoy quicker and more reliable response times.

 

A mirror site in Europe, and later perhaps in Asia, would be a major enhancement. And in addition to that, it would cut down the discussions about the unreliability of geocaching.com servers and people would stop thinking about creating their own local cache listing services.

 

Jeremy, have you thought about this solution?

 

Regards, Captain Morgan

[+New England n00b]

1-2 mirror sites around the globe

 

Wouldn't that make it hard to read - I mean, all backwards and stuff....?

 

 

It would be interesting to see how they can split the database. When they sycnhed them so that everyone gets the same data, would it require the site to go down for a short period? And at what frequency would teh synching be?

 

Interesting question (mirrors), interesting problems... curious to see what TPTB think of such a thing (though it wouldn't likely affect me in the USA).

[+Divine]

Interesting question (mirrors), interesting problems... curious to see what TPTB think of such a thing (though it wouldn't likely affect me in the USA).

It could affect you while you're in the USA. If the US server went down, the same data would be available on the mirror server for you too.

[+New England n00b]

That's a good point. Then the question is, where would the hardware be placed, and who would maintain it, if not the current PTB?

 

They would have to be employees, I would think, rather than volunteers (and all the international law headaches that go with that).

 

What if they had a spat with the US PTB, and decided to create problems of some sort. (Think along the lines of The "Litter"box). Problems I might see are intentional db corruption then synched.... exposing teh raw db to the world+dog... whatever.

 

Of course these are unlikely problems, but they are questions that TPTB must ask themselves. And there is the $$$ issue behind it all.

 

To be clear, I think a mirror is a good idea - I'm not knocking it at all. Just thinking out loud...

[+Captain Morgan]

It would be interesting to see how they can split the database. When they sycnhed them so that everyone gets the same data, would it require the site to go down for a short period? And at what frequency would teh synching be?

Naturally those sites should be online mirror / replica sites. It should be possible to build such a system, though it's more difficult than regular mirror.

[+Captain Morgan]

That's a good point.  Then the question is, where would the hardware be placed, and who would maintain it, if not the current PTB?

 

They would have to be employees, I would think, rather than volunteers (and all the international law headaches that go with that).

No, the servers could be in some server hotell, so there's no need for more employees.

Edited September 30, 2004 by Captain Morgan

[+Divine]

Those all are very valid points n00b. As long as I'm concerned, the mirror server could by all means lie on the US turf too, at least for starters. The point is to keep GC.com available 24/7, and if the Seattle machine goes down, the mirror server(s), wherever they are, could provide the same information. Of course, in the long run it would be more advantageous to have mirrors on different continents.

[+Mopar]

The problem as I see it is the data on GC.com is VERY dynamic. How often would these mirrors sync? Once a day? Twice a day?

Could you imagine all the complaints if people logged a cache and it didn't show online for 24hrs? What if someone needs to move a cache, or disable/archive one?

New caches. FTF is very popular in some areas/groups of cachers. If new caches took 12-24hrs to propagate to the mirrors, many people would still stick with the main site to get the freshest data. People in other countries would complain they are "second class citizens" because they are delayed in getting the new caches. I guess to be fair all new logs/cache changes would go into a queue, and not be posted until the the next database sync. Would people be happy about submitting a log and not having it show up on the page for 12hrs?

What about the forums? What a mess THEY would be if the mirrors only synced a few times a day!

 

The only solution to the stale data would be that the databases would have to sync more often. What would be "often enough"? Every hour? Every 30 minutes? Every 15 minutes? Somehow I don't think things around here would go faster if 2 remote copies of the database had to be synced every few minutes.

 

The stated reason for this enhancement is to ensure 100% uptime of the site.

First off, the site rarely goes down, and even rarer for it to be down for more then a few minutes. If it goes down, even for a few hours, SO WHAT? This is a game, not an air traffic control system. Nobody is going to die or get injured because the site is down. Nobody's company is losing money or productivity because the site is down (probably quite the opposite! ). Surely there is SOMETHING else to do with your life for an hour if the site is down?

 

Do we really need to address all the headaches of mirror sites to solve a problem that shouldn't even BE a problem?

Edited September 30, 2004 by Mopar

[CoyoteRed]

Issues of lag are irrelevant if you set up the scheme so folks use a more local server.

 

Say you just split the US into two areas, the East and the West. You encourage folks in the east to use the server that is in the east and vice versa. You would still sync the data between the two, but you would only see a lag from the other area.

 

I don't know about you but I just about could care less about the caches in the west. I only care about the caches near me. In other words, for me, lag doesn't even begin to come into play.

 

The only folks that would be effected by lag are those who would have a caching range in both areas and those whose areas over lap theirs.

 

To prevent a ripple effect of logs being logged on a machine outside that cache's area, you strongly encourage logging certain caches on one machine unless there is a good reason not to--like the local machine being down. Simply make the dividing line something concrete like A, B, and D state log on one machine while the others log on the other one.

 

The problem with lag would be a bare minimum.

 

A second scheme might be to have diverse mirrors for outgoing data and single point handling incoming data most of the time. I can only imagine that the amount of data going out is many times larger than the amount of data coming in. I don't know enough about IT to know if this is workable, but it's a thought. If the main point goes down, the mirrors can hang onto the data until the main point it back up. Being quite simply, many times I don't care if the data is up-to-date, I just want to be able to write a log, send it, and move on to the next one. AND NOT wait to see if the site took it or tossed it aside. In this case "lag" can be thought of as "buffering."

 

...people would stop thinking about creating their own local cache listing services.

Nothing will stop people from talking about their own listing service without TPTB changing their business model in a very substantial way.

 

EDIT TO ADD: While mirrors might be a workable solution. The best solution would probably just get the proper hardware in to get the job done and, if necessary, move to another provider.

Edited September 30, 2004 by CoyoteRed

[+Divine]

If it goes down, even for a few hours, SO WHAT? This is a game, not an air traffic control system. Nobody is going to die or get injured because the site is down. Nobody's company is losing money or productivity because the site is down (probably quite the opposite! ). Surely there is SOMETHING else to do with your life for an hour if the site is down? 

 

Do we really need to address all the headaches of mirror sites to solve a problem that shouldn't even BE a problem?

Yes, there are plenty of things I could do when I'm not accessing GC.com, thanks for asking. While you might have a field day with other activities while the site is down, some people still prefer it to be accessible whenever they choose to access it. It's not about the lack of things to do, it's about having a service for which a bunch of people has paid for and the service providers willing to have the service operational all the time for their customers.

 

I know the site isn't totally down 'very often', and the importance of the site isn't comparable to air traffic control, but last time I checked this is GC.com website forum, where we can discuss new features (or feature requests), bugs, etc. specific to the Geocaching.com site. So pondering solutions to existing - if even small - problems should be on-topic even if we could do something else with our lives when the problems occur.

 

Now, I don't know practically anything about the mirroring techniques or the investments and workload needed for the job, and it might well turn out to be too unstable/expensive/not-worth-the-trouble/whatnot, but as a regular internet user I've seen smooth mirrored sites working perfectly well, so I'm very interesting to hear about the possibilites of having it done here. Or if not mirroring, then other plans to provide a stable 24/7 service.

[+Captain Morgan]

Although I work with computers and networks, I don't know much about remote mirroring or dynamic replica servers. But I'm quite positive that the system could be built anyway. I think that for example VISA has such a system. Ok, they have much more money to spend, but I just wonder how much the creating of that kind of system would cost.

 

Here's two links to quite similar solutions, but I'm not sure if they fit to geocaching.com:

http://www-ece.rice.edu/~sranjan/publications/ward-hp.ppt

http://www.microsoft.com/technet/prodtechn...oy/hasog05.mspx

 

Another solution for 24/7/365 services could be locally mirrored servers, where the load is shared equally between them. In that case there should be also a backup route to internet, so if the normal route goes down, servers could use the backup route.

 

I agree that the site goes totally down rarely, but somehow it seems to go down always when there's a night in USA and a day in Europe. Of course I could do something else when the site is down, but hey, there's a lot of paying members and if I understand right, Groundspeak would like to have them more. And if people pay for services, they have a right to ask it to work also.

 

Big problem is also that the hardware of geocaching.com does not seem to be able to handle all the load it receives, especially during Sundays and Mondays. That makes 104 days per year when the site is almost down. If the servers are dynamically mirrored, even locally, and load shared between them, it would help a lot.

Edited October 1, 2004 by Captain Morgan

[+Captain Morgan]

What about the forums? What a mess THEY would be if the mirrors only synced a few times a day!

IMO, there's no need to make a replica of forums, for me it does not matter if forums are down even a week or two - I can live without them. But the main issue is the reliability of the main site - I can't live without geocaching

[da_snoop]

You may like to refer to this topic in the forums: New Servers for Geocaching.com

[+CompuCash]

You may like to refer to this topic in the forums: New Servers for Geocaching.com

thank you - beat me to it -