rrTV-PHOTO   New HD TV
HOME   rrTV-PHOTO   GALLERIES   MY GALLERY   HELP-FAQ
myHOME PM pmRR MEMBERS 791 ONLINE 25 EVENTS SEARCH REGISTER  START HERE
 
3 pages [ <<    <     1     ( 2 )     3     NEXT    >> ]2597 viewsPOST REPLY
E-flite . Next D . Fast Lad Performance

.
.
Main Discussion > Welcome to the rrBackup server
 
 
Drunk Monk
rrProfessor
Location: Preston, UK

I am alittle supprised that backups aren't done more regularly. When I was running my gaming site I used to back the db up hourly with a mysql dump and email it to myself (my server was in the usa) and it was a huge db with 20K members and forum. I emailed it to a gmail account since they give plenty of storage and just download to my pc once a day. You can never have too many backups and clear out the crap once a month :P

Only problem with this method is that it can take along time to restore a big complex db unless you have hands on the server.


Stephen

I only open my mouth to change feet.....
05-27-2008 10:33 AM
 
 
TachyonDriver
Veteran
Location: Chipping, Lancs, UK

At least I'd managed to get my long post in on the hot girls thread

A shame my diatribe on spending 3 trillion didn't survive - I was quite proud of that!

Tach.

Little Spinning Bundle of Joy® DON'T DISS THE DINO!!
05-27-2008 11:18 AM
 
 
JAGNZ
rrProfessor
Location: Auckland, New Zealand

Yeah I had written a few pearls myself. Se la vie.

Jason Greenwood

Fury Extreme, T700N, 2X T600N's, Trex 500, DSX9 2.4
www.3dheli.co.nz
05-27-2008 12:16 PM
HOMEPAGE  
 
 
tchavei
rrProfessor
Location: Portugal

Although RAID is widely used in the world, I personally stopped using the system (raid 5 that is). The only times I lost data big time was with raid servers.

Disks have a normal lifespan. To have a great system, we usually buy 5 or more discs from the same batch to improove performance. After a few years when one goes down, you can be sure the next one will go down shortly after. Its pretty common for disks from the same batch to die together. I had a big system with 1.5K users where one of the raid disks died at 4 am. I had the server programmed to imediately use the spare slot and rebuild the array... well 2 hours later disk #4 died as well and all data went down the drain. Of course I had backups from the day before but still 12h of activity was lost.

What is even worse is that even if you manage to rebuild the array, 7 years after initial purchase, its almost impossible to find the same type of disk (SCSI) and at least my Raid controllers really hate when I start swapping disk sizes and geometries.

Since that fiasco I'm using Raid 1 (mirror) with inexpensive Sata disks. If one goes down, the other has a perfect copy and you can plug in a new one (different brand and size) without any ill effects.

my two cents
Tony


--------------------
"Perfection and patience usually walk side by side..."
05-27-2008 12:54 PM
HOMEPAGE  
 
 
Nathan
Senior Heliman
Location: Elkhart, Indiana

Quote 
IM not up with computers...... but am willing to learnwats a server, wats its purpose?

A server brings your steak to the table! Then you give her a tip.


Nathan
05-27-2008 12:54 PM
 
 
tchavei
rrProfessor
Location: Portugal

Sorry, didn't saw the question...

Quote 
IM not up with computers...... but am willing to learnwats a server, wats its purpose?

A server does just that... it serves resources to other computers (users) on a network (intra or inter). A server can fullfill many purposes as collecting/sharing/storing data, manage users and accounts, etc.

Usually we call a server any computer whos sole function is to provide a service to other computers/users. If you share your printer with another computer on your home network, your computer is acting as a server to that computer. If you share your internet connection so that other computers at home connect through you, then you are the internet server for those computers. If you have a router and don't need any particular computer to be ON so others can access the internet, then you don't have a server.

Tony


--------------------
"Perfection and patience usually walk side by side..."
05-27-2008 01:05 PM
HOMEPAGE  
 
 
Pistol_Pete
Elite Veteran
Location: Tampa Bay non-Buccaneer

Quote 
Seven years is a good, long life.

MLBF for the average PC is 5 years due to hard disk.

concur with RAID 1 comment...best of both worlds in my book.

<><>...the lunatic is in my head...<><>
05-27-2008 01:13 PM
HOMEPAGE  
 
 
rcadd1ct
Elite Veteran
Location: Richardson, Texas

Depending on server load, you might want to look into VMWare ESX server.

It lets you have virtual machines on physical HW.

If you make a cluster group, they can move VM's between the physical servers.

At our office we have 186 VM's on eight physical servers.

The physical servers are nothing to sneeze at, but maintenance is easy.

Move the VM's with vmotion, it loses one ping packet as it moves, then shut down the physical box to do maintenance. Move onto the next for a rolling upgrade.

-RCA .......... Making Cuisinarts Fly!!!!!!!
05-27-2008 01:18 PM
 
 
tadawson
Elite Veteran
Location: Lewisville, TX

For high availability solutions, there are far better solutions than VMWare. VM is great for test environments and things that don't need the full "grunt" of a single box. For HA, solutions like PolyServe or are far superior, and can actually allow parallel processing on a lot of apps. Using a quality database with data integrity checking and recovery goes a long way too - if a system is properly designed, more than one backup a day is excessive and should not be needed.

Oh, and while I think they upgraded it beyond this issue (or plan to), VMWare's VMotion has (had?) a pretty tough limitation - both systems need to be up to get VMotion to work. IE, if one system dies, VMotion can't do anything to help . . . it's more of a load balancing tool than an HA tool . . .

And regarding RAID, when you get to "real" RAID - IE, not the embedded junk that comes with servers, but rather a real, dedicated, fully redundant system, RAID-5 (and now RAID-6, like 5 with dual parity) pretty much rules the market - RAID-1 has pretty much been abandoned in larger installs due to the wasted disk space, and having no additional resiliency over RAID-5 (or 6). For what it's worth, one of the main "pros" of RAID-6 is the scenario that tchavei described - secondary drive failures on rebuilds - the dual parity gets you over that hump. That, and using commercial quality disks and not the "home pc" junk makes a lot of difference in reliability as well. They cost more, but live a LOT longer . . .

- Tim

The more I touch electrics, the more I grow to hate them . . . .
05-27-2008 05:07 PM
 
 
tchavei
rrProfessor
Location: Portugal

the $15K server where that disaster happen did had a dedicated scsi controller and hotswap unit...

Tony


--------------------
"Perfection and patience usually walk side by side..."
05-27-2008 06:02 PM
HOMEPAGE  
 
 
tadawson
Elite Veteran
Location: Lewisville, TX

Most of the dedicated RAID that I work with is $100K and up, so we are still talking seriously low end on your stuff . . . and that price does not include any servers - the RAID is completely external.

- Tim

The more I touch electrics, the more I grow to hate them . . . .
05-27-2008 06:22 PM
 
 
HeliSmith
Senior Heliman
Location: Canyon Country, CA

That's cool Mark let us know. I had a post disappear and couldn't figure out if I was dreaming of posting on Run Ryder too. lol

*** Team HeliHobby.com ***
05-27-2008 08:45 PM
HOMEPAGE  
 
 
crofty
Senior Heliman
Location: Phoenix, AZ - USA

Availability is a balancing act. I design high availability software/hardware architectures for a blue chip global financial company. I am painfully aware of the cost of creating highly available systems. But this is RunRyder. I'd rather Mark spend his time and money developing the content and interfaces. I feel sorry for Mark that this happened twice in such a short time but that's the risk downside of the balancing act.

Why should he pay for managing hourly backups, extra redundancy etc for the one in a million condition that happened this week. It's not worth it. (If it had been my bank account now that's different.)

This thread reminds me of a day at the field. Whenever there is a problem with my heli there is always a flurry of suggestions on how to fix it. Some are useful some are not but nine times out of ten you just want a bit of quiet time to look at your heli and work it out for yourself without a bunch of people yelling suggestions at you telling you stuff you already know, can't afford or have already tried.

So far the site has been more available than any of my helis! Chin up Mark, your doing a great job and thanks for the service that has helped me and so many others.

I have nerves of steel but thumbs of jelly!
05-27-2008 11:31 PM
 
 
legoman67
Elite Veteran
Location: Nanoose Bay B.C, Canada

so basically if this server fries the whole site is toast? uh oh

Matt M.
http://www.filepile.ca
05-28-2008 12:41 AM
HOMEPAGE  
 
 
Mark Ryder
Elite Veteran
Location: Encino, CA

Yesterday the server fried. RR was back online in less than two hours. Minus a half a days posts.

Mark Ryder - 818-996-2222 - RunRyder Dot Com LLC
05-28-2008 12:44 AM
HOMEPAGE  
 
 
JAGNZ
rrProfessor
Location: Auckland, New Zealand

I guess 3 outtages in 3 days with a lot of lost posts each time might cause you to look at a different database/server option.

Jason Greenwood

Fury Extreme, T700N, 2X T600N's, Trex 500, DSX9 2.4
www.3dheli.co.nz
05-28-2008 12:19 PM
HOMEPAGE  
 
 
tchavei
rrProfessor
Location: Portugal

Quote 
Most of the dedicated RAID that I work with is $100K and up, so we are still talking seriously low end on your stuff . . .

We work in different worlds mate

Tony


--------------------
"Perfection and patience usually walk side by side..."
05-28-2008 12:27 PM
HOMEPAGE  
 
 
TrexRookie
Key Veteran
Location: San Francisco, CA

Quote 
The RR front end server was due for upgrade.

It's glitched twice in the past week so time to say good bye to it.

he is... we're currently on a backup server.. keyword... backup.
05-28-2008 12:28 PM
 
 
rcadd1ct
Elite Veteran
Location: Richardson, Texas

I am not sure what version of VMWare you looked at.

With HA/DRS if a host dies, you can have another host pick up the lost server and run in.

Yes, it is a reboot, but the data will be there, and the downtime is a virtual reboot.

-RCA .......... Making Cuisinarts Fly!!!!!!!
05-28-2008 12:59 PM
 
 
del19_82
Heliman
Location: Glasgow, Scotland

Is the server likely to drop again??

www.waterfoothelis.com
05-28-2008 04:28 PM
HOMEPAGE  
 
 
3 pages [ <<    <     1     ( 2 )     3     NEXT    >> ]2597 viewsPOST REPLY
Ace Hobby . Esprit Model . Thunder Power RC

.
.
Main Discussion > Welcome to the rrBackup server
 PRINT TOPIC Advertisers 

Subscribe to This Topic

Thursday, January 8 - 10:35 pm - Copyright © 2000 - 2009 runryder.com | email | link to rr | runryder needs cookie