[Hcoop-discuss] Next hardware configuration/Email service

[Justin S. Leitgeb]

Sounds great.  I would be interested in contributing a bit more to get 
us started as well.

I guess the tough part is figuring out an architecture that is robust, 
while still allowing us to provide "generic" hosting services.  What 
kind of redundancy do you think is reasonable?  Obviously we could fill 
up a rack pretty quickly depending on the redundancy that we want to 
provide.  I know that you mentioned a robust file server and it seems 
like this could help us out, but in practice I haven't seen web servers 
hitting a network filesystem.  I'm worried about performance issues and 
scaling here (I haven't tried this but there are some threads on the web 
about performance and scaling with Apache over a network file system 
that concerned me).  Might work with something like RAID 10, but I know 
applications like IMAP work horribly over NFS and AFS.

An alternative would be to set up a couple of web nodes, and give users 
an account on one or the other.  Both could have RAID 1, or no disk 
redundancy.  We could replicate between the two (e.g., rsync) in order 
to recover quickly in the event of a failure, and MySQL replication 
could work from one to the other, adding cheap DB redundancy.  Later it 
would be nice to have a database cluster, perhaps a 4 node MySQL 
configuration that would be suitable for dynamic web-based applications, 
but obviously this can wait.  Then it seems we should still need to have 
a third system for services such as IMAP, etc., that we don't want to 
give normal users access to.  Since this would be a possible single 
point of failure, it should have a high level of RAID, power redundancy, 
etc.  Perhaps RAID 10 on this system because IMAP is so I/O intensive.  
A hardware firewall would be a nice addition as well.  This would be a 
3-node starting configuration.

Specifically, what kind of architecture were you thinking about?

Justin

Adam Chlipala wrote:

>Justin S. Leitgeb wrote:
>
>  
>
>>The high cost that I estimated for setting up multiple servers was 
>>really what motivated me to ask about co-location before hardware or 
>>specific configuration details.  I couldn't imagine the co-op coming up 
>>with more than what would be needed for a single, solid 1U server.  
>>Specifically I was thinking that we could build up from something like a 
>>Dell Poweredge 1850, which is 1U and would set us back at least $3000 
>>per unit with any kind of hardware RAID (this would give us RAID 1 with 
>>2 300 GB disks, but only 1 CPU).
>> 
>>
>>    
>>
>There are two main classes of reasons why this could be feasible.
>
>First, as Nathan suggested, some of us are more willing to spend money 
>on HCoop than you might realize.  I wouldn't be averse to throwing a 
>four-digit amount into the pot, especially if the idea was that member 
>dues would pay me back relatively quickly.  Other people have much more 
>expensive hobbies. ;-)
>
>There's also the possibility that our colocation provider would sell us 
>servers on an installment plan.  This is essentially what you are 
>getting with dedicated server plans, except that there you don't get to 
>own the server after a particular number of installments.
>
>  
>
>>Although I think it would be great if hcoop could buy a rack full of 
>>servers within the next year, I don't think that this is necessary, or 
>>even the best way to grow, at least financially.  We should be able to 
>>move in a positive direction one server at a time for now.  We can start 
>>by putting services on other boxes as we have resources to do so.  With 
>>one 1850 I was thinking that we could do something like moving mail 
>>(IMAP and sendmail) off of fyodor onto the new server.  Services like 
>>IMAP shouldn't be on shared servers anyway, so this would be a healthy 
>>move both for hcoop members who use email and those who don't rely on it 
>>so much.
>> 
>>
>>    
>>
>I probably haven't been explicit enough about my thoughts on our current 
>set-up.
>
>I already said that we configured Abulafia in a time of ignorance, and 
>so it isn't a good part of a solid hosting set-up.
>
>We also have issues with fyodor.  We don't own it, so we can only use it 
>at InterServer.  InterServer essentially tricked us into signing up with 
>them with false advertising; whether or not they did so purposely 
>doesn't change the consequences.  They offered an on-site backup service 
>and then refused to sell it to us, despite having a page up for a few 
>weeks saying that ordering would resume in a day.  We learned this after 
>signing up and spending two months getting our rented server ready, so 
>we decided to stick with them in the short term for pragmatic reasons, 
>but the lack of off-machine back-up services is a serious concern.
>
>As a result, I'd like to sever all relationships with InterServer as 
>soon as possible.  An incremental scheme of moving some services to 
>other servers wouldn't satisfy me so much; I'd know that all data kept 
>on fyodor was in very serious danger of being lost, as it is now.  Like 
>my current conception of Abulafia's role, I'd instead propose keeping 
>fyodor running as a backup during the transition period and then getting 
>rid of it.
>
>  
>
>> The cost for a configuration like this could be offset by 
>>opening up hcoop services to a new population.  Searching the web for a 
>>minute I didn't find any co-operative email providers, and I think that 
>>this would be a great thing for individuals interested in more 
>>socially-conscious alternatives to gmail.  We could use IMAP, 
>>squirrelmail, and a few small custom programs to open up email services 
>>to non-technical users for a small fee.
>>
>>    
>>
>We've been interested in a while in designing service offerings to 
>appeal to non-technically-savvy people.  Simple web hosting without 
>log-in access is another service in this niche.  I've been recommending 
>that we wait until we have a solid infrastructure that competes with the 
>best commercial offerings.  My main reason is that non-geeks will react 
>more violently to small hitches, and those are certain to come up as we 
>work towards our ideal set-up.  Also, I'm hoping to move everyone to our 
>new set of servers shortly after they are ready, and non-geeks would 
>probably find this migration process to be arduous instead of trivial if 
>bothersome.
>
>I have hopes that our next set-up (in its first incarnation, not after 
>later additions of new servers) can meet the criteria needed here, the 
>main ones being back-ups and redundancy.  A large data loss incident 
>would certainly not help our image, and the more members at the time and 
>the less technically savvy they are, the worse it would be.  This is why 
>I want to create a complete redundant infrastructure from scratch.
>
>_______________________________________________
>Hcoop-discuss mailing list
>Hcoop-discuss at hcoop.net
>http://hcoop.net/cgi-bin/mailman/listinfo/hcoop-discuss
>  
>