The Time Has Come: Getting a new server
- Thread starter Mike Bobbitt
- Start date
- Reaction score
- 5,973
- Points
- 1,260
Flavus101 said:
It is, but I would worry about classifying us by our ability to send money rather than by our willingness to help it all the ways we can: most important by being Moderators, also by making good, helpful posts and also, indeed, by sending money when it's needed ... I think it's important that Moderators are clearly visible as Directing Staff, I think we can all identify the regular participants in discussions - some of us are members, some veterans and a few are even fixtures or legends
, and I, personally, think it is sufficient to show that some people can and do support the site. I'm not sure making some more 'important' because of how much they send compared to others would achieve much.
- Reaction score
- 0
- Points
- 210
E.R. Campbell said:It is, but I would worry about classifying us by our ability to send money rather than by our willingness to help it all the ways we can: most important by being Moderators, also by making good, helpful posts and also, indeed, by sending money when it's needed ... I think it's important that Moderators are clearly visible as Directing Staff, I think we can all identify the regular participants in discussions - some of us are members, some veterans and a few are even fixtures or legends
By having a separate "Builder" subscriber that is more or less creating a tier above the current subscriber, no? I agree with you completely on being cautious, we have to be careful with how much we label people through monetary donations. I think that is why there is a set subscriber rate and then if people feel as if they wish to give more to the site they can use the donate feature. (Correct me if I am wrong, I am not exactly sure how that system works).
You now have another subscriber Mike! Army.ca has been a huge help to me and it has been a very enjoyable time.
- Reaction score
- 198
- Points
- 1,210
I've thought about multiple subscription tiers in the past, but going the other way. The thought is to offer a $10 tier that does not provide the goodies (shirt, badge), but does allow the user to support the site and tout the Subscriber badge. The general idea there is to make it more widely accessible. I still think this may be viable; when we are in need, it seems the majority of those who provide support are the same ones that did the last time (and the time before). That's not a bad thing, but I hate to keep leaning on the same group over and over.
Maybe I've had too much exposure to sales folks, but their notion of prioritizing "new business" makes some sense. You can't keep milking the same crowd and expect to grow. You need to expand, even if by small amounts.
Moving to Linode is definitely more work than moving to a new EC2 server. In the latter case, I literally image, restore and assign our IP address to the new more powerful server. For Linode, it's more of a "build it piece by piece in parallel until it looks the same" scenario. The cutover process involves propagation delays that can be measured in days in some cases. The actual difference in my time is probably 3-6 hours vs. 3-6 days, and in downtime it would be hours vs. days as well. (Your mileage may vary, of course, depending on caching, etc.)
Although not identified as a factor in previous posts, cutover time has definitely been in the back of my mind.
Maybe I've had too much exposure to sales folks, but their notion of prioritizing "new business" makes some sense. You can't keep milking the same crowd and expect to grow. You need to expand, even if by small amounts.
Flavus101 said:
Moving to Linode is definitely more work than moving to a new EC2 server. In the latter case, I literally image, restore and assign our IP address to the new more powerful server. For Linode, it's more of a "build it piece by piece in parallel until it looks the same" scenario. The cutover process involves propagation delays that can be measured in days in some cases. The actual difference in my time is probably 3-6 hours vs. 3-6 days, and in downtime it would be hours vs. days as well. (Your mileage may vary, of course, depending on caching, etc.)
Although not identified as a factor in previous posts, cutover time has definitely been in the back of my mind.
- Reaction score
- 0
- Points
- 210
Mike Bobbitt said:
Thanks for that! Do you feel the Linode server would be able to handle our current traffic and provide enough growth for the rest of the year if things stay on the current pace?
What I liken this to is buying a house. Let's say you want to get a nice house for your wife and you, this means you will need enough room for two people. Eventually you have a family and that means you need room for 4. The kids move out and now you have a bunch of wasted space...
The problem I see with getting a three year term is that a lot can change in three years. Growth may be exponential and we run out of room again a year and a half in. Growth could die down and we are now running a server that is 3x more powerful than we need for a lot more money. My train of thought is to eventually get to this m3.2xlarge server, but to have a sort of bridge server in between. If we can run the Linode for a number of months until it maxes out than switch to the Amazon we would be using money more effectively. The biggest question is how long we can run on the Linode before needing to switch to a bigger server.
E.R. Campbell said:It is, but I would worry about classifying us by our ability to send money rather than by our willingness to help it all the ways we can: most important by being Moderators, also by making good, helpful posts and also, indeed, by sending money when it's needed ... I think it's important that Moderators are clearly visible as Directing Staff, I think we can all identify the regular participants in discussions - some of us are members, some veterans and a few are even fixtures or legends
I don't think there is any need to identify those that build, other than maybe a note somewhere pointing out that there are xxxx number of members and subscribers who belong to that group with thanks. Personally identifying someone is just not on.
- Reaction score
- 3,106
- Points
- 1,160
A little something sent, get yourself something nice.
- Reaction score
- 1,401
- Points
- 1,160
Mike,
Interac e-Transfer sent this a.m.
Thank-you for all you do.
Interac e-Transfer sent this a.m.
Thank-you for all you do.
- Reaction score
- 198
- Points
- 1,210
Folks, a quick update on this. I fell behind due to unexpected events, but we are back on track. I tentatively plan to perform the upgrade tomorrow morning, assuming I can get everything lined up correctly.
Colin Parkinson
Army.ca Myth
- Reaction score
- 11,930
- Points
- 1,160
Mike there is no reason to apologize, the fact that you are working on this at all right now is deeply appreciated.
- Reaction score
- 2,849
- Points
- 940
I definitely prefer the "Go grab a coffee" screen to the "database error" one :nod:
George Wallace
Army.ca Dinosaur
- Reaction score
- 184
- Points
- 710
Journeyman said:
Would make a nice screensaver....... ;D
- Reaction score
- 198
- Points
- 1,210
OK, migration is complete. Any coffee you are required to take from here on out is voluntary. 
Please let me know if you have any problems or see anything that just doesn't quite look right.
Cheers
Mike
Please let me know if you have any problems or see anything that just doesn't quite look right.
Cheers
Mike
- Reaction score
- 198
- Points
- 1,210
If you need a break any time, it's available at http://coffeebreak.army.ca
- Reaction score
- 6,080
- Points
- 1,160
Thanks Mike, especially at this time. 
Now, let's hope it does the trick for a long while.
Great job.
Now, let's hope it does the trick for a long while.
Great job.
- Reaction score
- 198
- Points
- 1,210
I've been watching things in the background, and there is a definite performance improvement. It's going to take some fine tuning on my end to get things right, but it is definitely an improvement.
I'll outline the changes, since they are "none of the above" as listed previously.
Instead of throwing more power behind the existing server (which handled everything), I decided to split out the database onto a dedicated server. So Army.ca now runs over 2 servers, one (the front-end) handling the web server, scripts, content, etc. The new server (db.m3.large) is dedicated to storing and managing all the data.
So when you make a post, you invoke a script on our front-end server, which captures the text and stores the actual content on the back-end server. When someone else comes to read that post, essentially the same process happens again, this time to fetch the data.
This division of duties allows each server to focus on it's task only, and when one gets bogged down, it doesn't affect the other. A side-effect of this is that the current server has not been replaced or upgraded at all. It just has less to do now. This means that the original "reserved instance" (which you will recall is paying a fee up front to pay less over time) expires on 2015-05-03. I've also purchased a reserved instance for the new back-end server, which now expires on 2015-06-12.
This secures our place for the next year, but allows us also to expand the capabilities or one or both servers in a year's time.
That is, it should fit our needs now but doesn't lock us in and prevent further growth.
Thanks again for your support and patience through this, it would not have been possible otherwise.
Cheers
Mike
I'll outline the changes, since they are "none of the above" as listed previously.
Instead of throwing more power behind the existing server (which handled everything), I decided to split out the database onto a dedicated server. So Army.ca now runs over 2 servers, one (the front-end) handling the web server, scripts, content, etc. The new server (db.m3.large) is dedicated to storing and managing all the data.
So when you make a post, you invoke a script on our front-end server, which captures the text and stores the actual content on the back-end server. When someone else comes to read that post, essentially the same process happens again, this time to fetch the data.
This division of duties allows each server to focus on it's task only, and when one gets bogged down, it doesn't affect the other. A side-effect of this is that the current server has not been replaced or upgraded at all. It just has less to do now. This means that the original "reserved instance" (which you will recall is paying a fee up front to pay less over time) expires on 2015-05-03. I've also purchased a reserved instance for the new back-end server, which now expires on 2015-06-12.
This secures our place for the next year, but allows us also to expand the capabilities or one or both servers in a year's time.
That is, it should fit our needs now but doesn't lock us in and prevent further growth.
Thanks again for your support and patience through this, it would not have been possible otherwise.
Cheers
Mike
Good............now you have about 8-9 months to set up a build fund of say.....$50.00 each. This will allow you to go into the next renewal with $$ in hand, whether it is simply a renewal of what we have or an upgrade.
What you don't use of the annual builder's fee (voluntary) gets put aside to accumulate for the next year. The site is growing, as are the costs. You are gonna need it.
Just to be clear.......the builders fee is voluntary and in addition to the subscription fee.
It's the cost of one short pub crawl.
What you don't use of the annual builder's fee (voluntary) gets put aside to accumulate for the next year. The site is growing, as are the costs. You are gonna need it.
Just to be clear.......the builders fee is voluntary and in addition to the subscription fee.
It's the cost of one short pub crawl.
Similar threads
