Building a server?

D

deadman_uk

[H]ard|Gawd
Joined
Jun 30, 2003
Messages
2,036
I am 19 and I am doing a 2 year HND in ICT. I have a new assignment where I have to build a computer system, listing full specifications, list software and alternative software and finding network components. I can either buy the parts seperately or buy a ready made package.

I no nothing about servers at all but this system must hold 100 000 customers and each customer must have 50mb of space.

My questions are..

1) (sorry my math is poor) How big of a hard drive should I buy if there are 100 000 customers who need 50mb of space each?

2) I know a fair bit about computer hardware so can't I just build a high spec PC and call that a server? (sorry I am a complete newbie when it comes to servers)

3) I will need some software for this, do servers require an OS, if so, I could use windows and install software like Microsoft Office? (again, I know nothing about servers)

Thanks for any help.
 
ok

100 000 users who need 50mb each

50mb x 10000 = 500 000mb

500000mb = 488.28125gb


Is this right?


I still need help with the other questions please
 
I think you need to check your math again.

Given the specs you've listed, it's quite difficult to determine what parts to put in the server. Just saying 100000 customers doesn't mean a thing. You haven't even stated what type of server it is supposed to be or what it will be used for.

As far as software goes, yes, a server requires an operating system. However, the OS and software needed depend largely on what the purpose of the server is. And most people aren't going to be typing up documents on a server, so office would be quite pointless.
 
100 000 users who need 50mb each <------- this, and


50mb x 10000 = 500 000mb <----- this, is it supposed to be 100,000 or 10,000

500000mb = 488.28125gb



If thats so, you need to stick another zero on the end.


3) I will need some software for this, do servers require an OS, if so, I could use windows and install software like Microsoft Office? (again, I know nothing about servers)

Yes, servers need operating systems. I recommend CentOs linux. Its a very stable, lightweight, and fast os. Less bloat then windows. Normal windows is not good for a multi user environment, however windows server 2003 is decent, but i still recommend linux over it.
 
Yeah lol I missed a zero out..

100 000 users who need 50mb each

 50mb x 100000 = 4.76837158 terabytes

I will list a small part of the assignment..

- Customers must have 50mb each and 24/7 access

- Computer system must cope with administration, customer access, files, customer web sites, email, billing etc

- Security of data and transactions is vital

- System must be expandable

- High speed printing facilities must be available for billing, logging etc

- LAN is needed

- Secuirty policy e.g passwords, encryption, physical sercurity

- Must list software and alternative software
 
Sounds like a bullshit assignment. Nobody in their right mind would use a single server for all those services, especially with so many users. Web and email at least should be on their own servers(and probably even multiples of each of those with appropriate load balancing equipment and such). Also, the 50megs/user, is that combined web and email usage, or just one or the other? As far as 24/7 access goes, you're going to want plenty of redundancy all around(at least in storage, power, and network links). All in all, this assignment sounds quite a bit more involved than I originally anticipated. Good luck.
 
Yea, this dosnt seem that possible to me. I work for a hosting company as a server admin / tech and we put around 300 people on the server before load issues become a factor. We use fairily powerful servers, dual xeons, so its not like we have inferior hardware. I dont think your going to be able to do this without a cluster setup, with hardware load balancing, and even then, your going to need a bunch of them. Not to mention the private networking backplane to mirror the information for backups to keep the 24/7...aw geeze the more i think about this the more my brain hurts...i gotta stop...
 
The 50mb per user is for email and webspace.

So I need multple servers? how many? and what is a cluster setup?
 
Sounds like the assignment is to design an ISP
You will probably need groups of servers for that number of users.
Like a group of file servers -> each machine handles X number of users
Then a group of email servers etc

Be sure to plan in some overhead.

A block diagram would be helpful. Break you big problem into a bunch of smaller pieces and work from there.
 
If I was going to do this for my work I would call a reputable company like HP for pre-built hardware. That way, if something goes wrong, you'll have the advantage of a single technical support line and awesome warranty coverage (we have a guaranteed 4 hour response time at my work). Buying server grade pre-built hardware ensures that as much redundancy as possible has already been built into the machine as possible (dual SCSI HD, dual PS, ECC mem, etc).

For the web and mail servers, I would recommend a quad proc Opteron system with the max of 4 GB of ram each (unless you use a 64 bit OS then you can buy as much RAM as you can afford).

For the print server, which sounds like it is going to be an internal use only server, I would recommend a dual proc Opteron box with again, 4 GB of Ram.

Each server should have at least 2 Gig NICs Teamed for fault tolerance.

You are required to have way more storage than a single server can realistically host. You can use multiple JBODS to reach your goal of 5TB of storage but would not be recommended. The biggest reason is that all that storage would have to be connected to a single server which would make it difficult to expand the environment and provide any sort of redundancy (the server dies, your screwed). I would recommend implementing a SAN. The initial cost of bringing in a SAN is high as you need to buy HBA's and a Fibre Channel switch, but the physical SAN cabinet is significantly less than buying a NAS. The SAN would provide all the data security, redundancy and scalability that you could ever want.

You can go Linux, but be aware that there are considerably fewer Linux admins out there than Windows admins and that they cost a lot more to employ. I would recommend Windows 2003 server 64 for all three systems. It is important to purchase a 64 bit OS because the next release of Exchange server will be 64 bit only. Going 64 bit up front will prepare you for the migration down the road.

Which brings up messaging; keep as many things MS as you can, it will simplify administration. Go with MS Exchange server. It is a relatively stable messaging platform with wide industry support.

Alternative software options would best be provided by qualified Linux professionals. You may also check out the Red Hat website to see if they have any products listed on there and there licensing requirements. Some will have a max DB size or a user limit where they will simply cease to function when that limit has been reached.

You should also plan a DR strategy. To protect 5+ TB of data you are going to need a pretty beefy backup system to be able to back everything up every night. Even if you choose a backup strategy like doing a full every Sat night and Incrementals during the week, you'll still have to address the Business Owners needs as what is an acceptable amount of downtime in the event of a critical failure and plan a backup strategy based on that as each will require a diff amount of time to restore. As far as backup hardware, you'll need a tape library of some sort. You'll also need to look at off-site storage locations to protect that data in the event that something happens to the DC itself.

After all that, you'll need to purchase the infrastructure (power, rack space, cooling, switches, cables, cable management, etc).

And don't forget that 100,000 users will require a significant amount of bandwidth to be able to access their data which means you'll proposal will need to include monthly costs of dedicated internet access. How much will be up to you and the Business Owners (in this case, your teacher).

I think that covers it. Probably overkill for this assignment but in a real world DC, this would be pretty standard.

If all that doesn't make your head hurt and your palms sweat, you may be in the right field.
 
wow huge post black, thanks so much! I don't understand alot of what you are saying though, you use alot of technical language/jargon.

If I was going to do this for my work I would call a reputable company like HP for pre-built hardware. That way, if something goes wrong, you'll have the advantage of a single technical support line and awesome warranty coverage (we have a guaranteed 4 hour response time at my work). Buying server grade pre-built hardware ensures that as much redundancy as possible has already been built into the machine as possible (dual SCSI HD, dual PS, ECC mem, etc).
Click to expand...
Good idea, I will probably select a few servers from the Dell website, I will say one of the reasons for choosing a pre made server would be technical support and warranty.

1a) What does redundancy mean? does it mean, protection via a backup?

1b) What is ECC memory?

1c) What is a dual PS? (dual power supply? which if one fails, the other kicks in?)

1d) What is a dual SCSI hard drive in relation to a standard one?

Each server should have at least 2 Gig NICs Teamed for fault tolerance.
Click to expand...
2) I think NIC is the network interface card, but I dont know much about them, is the 2gig on the nic the same as say the memory on a graphics card?

You are required to have way more storage than a single server can realistically host. You can use multiple JBODS to reach your goal of 5TB of storage but would not be recommended. The biggest reason is that all that storage would have to be connected to a single server which would make it difficult to expand the environment and provide any sort of redundancy (the server dies, your screwed). I would recommend implementing a SAN. The initial cost of bringing in a SAN is high as you need to buy HBA's and a Fibre Channel switch, but the physical SAN cabinet is significantly less than buying a NAS. The SAN would provide all the data security, redundancy and scalability that you could ever want.
Click to expand...
You use so much jargon here, I really cant make sense on what you are saying. I thank you for your time writing this, but this is a assignment which is suppose to be brief and basic.

3a) What are JBODS?

3b) You say to reach my 5tb target, I need multiple JBODS? well why cant I have a 3 Dell PowerEdge 2850 rack server for the webspace which supports 1.8TB internally and then if more space is needed, I could attach more hard drives externally to the servers?

3c) What is a SAN? and how would it provide data security?

3d) What are HBAs and NAS?
You should also plan a DR strategy. To protect 5+ TB of data you are going to need a pretty beefy backup system to be able to back everything up every night. Even if you choose a backup strategy like doing a full every Sat night and Incrementals during the week, you'll still have to address the Business Owners needs as what is an acceptable amount of downtime in the event of a critical failure and plan a backup strategy based on that as each will require a diff amount of time to restore. As far as backup hardware, you'll need a tape library of some sort. You'll also need to look at off-site storage locations to protect that data in the event that something happens to the DC itself.
Click to expand...
4a) What is a DR strategy?

4b) What is a DC?

4c) What is a tape libary?

thanks
 
I'll go ahead and answer the questions that I know enough about to answer (hopefully) without making a complete fool of myself.
deadman_uk said:
1a) What does redundancy mean? does it mean, protection via a backup?
1b) What is ECC memory?
1c) What is a dual PS? (dual power supply? which if one fails, the other kicks in?)
1d) What is a dual SCSI hard drive in relation to a standard one?
2) I think NIC is the network interface card, but I dont know much about them, is the 2gig on the nic the same as say the memory on a graphics card?
3a) What are JBODS?
3b) You say to reach my 5tb target, I need multiple JBODS? well why cant I have a 3 Dell PowerEdge 2850 rack server for the webspace which supports 1.8TB internally and then if more space is needed, I could attach more hard drives externally to the servers?
3c) What is a SAN? and how would it provide data security?
3d) What are HBAs and NAS?
4a) What is a DR strategy?
4b) What is a DC?
4c) What is a tape libary?
thanks
Click to expand...
1a) Straight out of the dictionary:
Electronics. Duplication or repetition of elements in electronic equipment to provide alternative functional channels in case of failure.
Click to expand...
1b) Error Correcting Code memory
1c) Basically. Except both are on simultaneously, if one fails, other keeps going. No downtime. Essentially a redundant PSU.
2) NIC = Network Interface Card. 2 gig NICs = Two Gigabit NICs (for redundancy). The "gig" refers to the speed of the cards. NICs dont have onboard memory.
3a) JBOD = just a bunch of disks. Unlike Raid arrays (except raid-0 which lacks redunancy), JBOD is basically a single drive spanned across multiple disks. No redundancy means if a drive fails, the data on it is lost (not a problem as long as data is backed up regularly), and downtime results. I would suggest using SCSI raid5 or raid6 arrays for the server holding the 50MB per customer. That way if a disk fails, no data is lost, and no downtime results. Keep in mind that RAID is not a backup solution, as it protects uptime, not data integrity.
3b) Multiple JBODs are a bad idea, as is adding hard drives externally for more space. JBODs and multiple external drives have no redundancy. You're not going to be able to achieve 24/7 uptime if a disk in the JBOD fails. And if the setup is supposed to support 100,000 clients with 50MB each, you're going to need multiple file servers with the data mirrored on each one, as well as hardware load balancing.
3d) NAS = Network Access Storage. Basically a box with hard drives in it, possibly with raid support, that you plug in to your network to have the data on it shared on the network.
4a) Disaster recovery strategy. What you're going to do to get back online if a "disaster" (eg. fire, power surge, etc) brings your setup offline.
 
Well, from my own experience this is what I can tell you. For just about 500 Users this is what is in the server room

1 Dell Poweredge for Mail
1 Dell Poweredge for Mechanicals

XServe and XServe RAID for all the files and FTP. The XServe RAID can hold a lot and it is great, I recommend the G5 XServe and XServe RAID, holds something like 7.4TB. If you want a PC, I definately recommend a Poweredge and some Seagate Drives. For Mail, you want Cheetah drives witht the 15.4K RPM because you want mail as fast as possible.


But TBH, setting up servers for 100,000 is one bullshit assignment, you need more then one server, you will need a couple. A computer handling 100,000 clients will either be so slow that it won't operate or will crash every few minutes. You really need at least 5 servers sharing the work.


Something like a dual socket dual core system can handle a good amount of work. Also the Quad G5's are great for servers, get the 10.4 Server OS and start.

My 000000.02 Cents.
 
Thanks for the replies! very greatful and thanks raven for explaining those terms.

ok I did a little learning a while ago, here is what I have come up with (this is the beginning of my assignment, its extremely rough and sketchy) by the way, i have a 2000 word limit on the assignment but things in my appendix page doesnt count towards my word count, i will list the specs in the appendix...

Introduction

I understand you are a new ISP company who plans to have 100,000 customers in total. You need a computer system that will cope with things like websites, email, billing, administration, and customer access.

I will be evaluating different systems and software saying whether or not they would be a good choice for your business.

A single system won’t be able to cope with all the tasks needed by the business so there will be multiple systems that will handle different tasks.

There will be

  • A system for customer websites
  • A system for printing
  • A system for email and customer access
  • A system for administration and billing
A system for customer websites

The plan is to have 100,000 customers who each have 50mb of storage each. 100,000 customers multiplied by 50 megabytes, equals approximately 5 terabytes of storage needed.

The Hardware

The server I have chosen is a Dell PowerEdge 2850 rack server. This server is great for handling customer websites because it is designed for networking to give customers the performance they want.

This server can support up to 1.8TB internally, however the more users a server has, the harder the server needs to work to keep up. A server can’t handle an infinite amount of users, there is a limit before the server can no longer cope.

The 100,000 customers will need to be split up onto separate servers to ensure the server can perform well.

17 Dell PowerEdge 2850 rack servers will need to be purchased and each will have a 300GB hard drive inside. 100,000 customers divided by 17 servers equals approximately 6000. This means there will be 6,000 customers per server which will guarantee that each server can cope well.

The servers will all be placed on top of each other in a rack. In the future, when more customers join, you can add more storage by adding more servers to the rack.


For a detailed list of specifications the Dell PowerEdge 2850 rack server has, please look in the appendix.

The Software

----

Backup

Backup is important so a storage area network (SAN) is needed to provide backup for the customer’s websites. I have chosen to use the Dell/EMC Fibre Channel CX500, It is designed for small to medium business and it can hold a total of 120 drives which is more than enough storage for 100,000 customers.

There will be 17 300GB hard drives inside which will provide the 5TB needed for 100,000 customers. If more customer space is needed in the future then you can add more hard drives.
Click to expand...
Does this make any sense to you guys? This is only a tiny part of my assignment but it doesnt have to be in alot of detail, and i dont need to list the latest hardware and software either, as long as ive attempted and made an effort, i will do well.
 
No offense, but that sounds like it was written by a 14-year-old for a 14-year-old. For example:
The plan is to have 100,000 customers who each have 50mb of storage each. 100,000 customers multiplied by 50 megabytes, equals approximately 5 terabytes of storage needed.
Click to expand...
Unless you are indeed writing this for a 14-year-old, take out the part where you say "100,000 customers multiplied..". For one thing, you say the same thing twice. Secondly, I'd hope that anyone starting a 100,000 customer ISP can do simple multiplication, so you might insult him by putting that in there. Instead, just say something like "Each of the 100,000 customers will have 50 megs for email/website, requiring a total of 5 terabytes of storage."

Seriously, though, who is the target for this proposal? Does this person have any technical experience, or are you supposed to assume he is technologically ignorant(in which case, why the hell is he starting an ISP with a goal of 100000 customers :D)? Also, are you not supposed to do anything regarding modems/telecom/networking equipment? That stuff is pretty essential to an ISP.

It's good that you have the SAN for backup, but you still don't have any storage redundancy, in either the servers or the backup equipment. Backing up data to a single hard drive is almost as bad as not backing it up at all. You really don't want to lose 6000 customers' data because a drive fails somewhere.
 
jpmkm said:
Seriously, though, who is the target for this proposal?
Click to expand...
According to his first post, this is an assignment for an ICT class.
 
No offense, but that sounds like it was written by a 14-year-old for a 14-year-old. For example: Quote:
The plan is to have 100,000 customers who each have 50mb of storage each. 100,000 customers multiplied by 50 megabytes, equals approximately 5 terabytes of storage needed.
Unless you are indeed writing this for a 14-year-old, take out the part where you say "100,000 customers multiplied..". For one thing, you say the same thing twice. Secondly, I'd hope that anyone starting a 100,000 customer ISP can do simple multiplication, so you might insult him by putting that in there. Instead, just say something like "Each of the 100,000 customers will have 50 megs for email/website, requiring a total of 5 terabytes of storage."

Seriously, though, who is the target for this proposal? Does this person have any technical experience, or are you supposed to assume he is technologically ignorant(in which case, why the hell is he starting an ISP with a goal of 100000 customers :D)? Also, are you not supposed to do anything regarding modems/telecom/networking equipment? That stuff is pretty essential to an ISP.

It's good that you have the SAN for backup, but you still don't have any storage redundancy, in either the servers or the backup equipment. Backing up data to a single hard drive is almost as bad as not backing it up at all. You really don't want to lose 6000 customers' data because a drive fails somewhere.
Click to expand...
No offense taken, but I am at the early stage, my sentances are in note form, I havent even checked the grammar or spelling. The teacher said explain how I got the 5TB storage so I dont see how else I can explain it.

I have to say how I will network everything together but I dont have to go into much detail. I dont have to list components for the networking or anything. It says in the assignment I need to draw a sketch..

·[font=&quot] [/font]a sketch and description of the proposed LAN (Workstations, interconnection devices, printers, storage, servers etc) including external connectivity to the Internet.


Can you help me out with the redundancy? What sort of redundancy would I need for both the servers and the backup SAN?
 
RavenD said:
According to his first post, this is an assignment for an ICT class.
Click to expand...
I have no idea what ICT is. And I'm assuming some fictitious guy at this fictitious ISP would be the target of this proposal(rather than his prof or whatever), so that's why I was asking what assumptions he could make as far as knowledge level and such.


Redundancy is pretty simple as far as servers are concerned. It's usually just a matter of implementing a RAID array. For basic redundancy, RAID1(mirroring) is usually sufficient. Dell's server configuration page should have a selection for RAID1. Then you just have to have two identical hard drives.

I'm not as familiar with ensuring redundancy for SANs. I'm sure your SAN device can support RAID, so it should just be a matter of doubling your hard drives and doing raid1 pairs like with the servers. But I'll let someone with more knowledge about SANs answer this.
 
http://www.englishforums.com/


*Your Friend*


Not my friend, but yours.
Click to expand...
Thanks, but I did say this is in note form. I only started this 1 hour ago (I got 2 weeks to complete it) and everything I am writing about servers I have to learn so if you have nothing good to say, please dont post in my threads.

I have no idea what ICT is. And I'm assuming some fictitious guy at this fictitious ISP would be the target of this proposal(rather than his prof or whatever), so that's why I was asking what assumptions he could make as far as knowledge level and such.

Redundancy is pretty simple as far as servers are concerned. It's usually just a matter of implementing a RAID array. For basic redundancy, RAID1(mirroring) is usually sufficient. Dell's server configuration page should have a selection for RAID1. Then you just have to have two identical hard drives.

I'm not as familiar with ensuring redundancy for SANs. I'm sure your SAN device can support RAID, so it should just be a matter of doubling your hard drives and doing raid1 pairs like with the servers. But I'll let someone with more knowledge about SANs answer this.
Click to expand...
ICT is Information Communication Technology. As far as I know, I can make many assumptions so no worries there.

I know a little bit about RAID, I had to do a small exam on RAID a few weeks ago. RAID 1 is writing data to 2 seperate locations so I need double the storage space as there is a 50% loss in storage.

I dont think mirroring would be a good idea because if the disk controller fails, you cant access your data. Duplexing would be a better choice since the mirrored data is connected to its own disk controller.

I dont know anything about RAID 6 but wouldnt RAID 10 be a good choice? it has the redundancy of RAID 1 and the speed of RAID 0.
 
You could get more indepth answers in the hard disk forum section regarding various RAID set ups. There were also threads there about RAID 6 + 10, etc. Do a search for it. :)

-J.
 
Ok I am really confused now.. let me explain

I need servers to hold 100 000 customers (each having 50mb webspace each).

Here is what I got ...

- 17 Dell PowerEdge 2850 rack servers

- Each of the 17 servers has a 300GB hard drive inside

- Each of the 17 servers also has another 300GB hard drive inside. Together the 2 hard drives per server are configured in a RAID 1 (duplexing) configuration to provide backup.

Now is this all the backup I need? How would I implement a SAN in here? What hardware do I need for a SAN? I tried looking at google but I am just confused.

If I connect my 17 servers to a Dell/EMC CX700 (http://www1.us.dell.com/content/pro...0?c=us&cs=555&l=en&s=biz&~tab=specstab#tabtop) then would this be my SAN?

Please help!!!
 
LOL, Sorry for all the acronyms. I forget the not everyone knows what they all mean. RavenD did a good job of explaining most of what I posted.


1a) What does redundancy mean? does it mean, protection via a backup?
-redundcany eliminates a single point of failure. For instance, if you have 2 Power Supplies (PS) in a server and one fails, the other will will be capable of running the entire server until you can replace the failed PS.

1b) What is ECC memory?
- Error Correcting mem. It doesn't actually correct memory errors but it is capable of detecing them and re-requesting that the info be sent. Using ECC mem is really a server standard as it prevents faulty data from being used/saved to the Hard Drive (HD).


1c) What is a dual PS? (dual power supply? which if one fails, the other kicks in?)
-See redundancy answer

1d) What is a dual SCSI hard drive in relation to a standard one?
-Most servers come with at least 2 SCSI HDs. When building the server you will be asked to configure your Array. Selecting RAID 1 when building the Array will provide HD redundancy for your server. If one drive fails, the other will be capable of running the server until you replace the failed HD. When you do, the Array Controller will re-build the array and restore the redundancy.

2) I think NIC is the network interface card, but I dont know much about them, is the 2gig on the nic the same as say the memory on a graphics card?
-yes, NIC refers to the network card. Unlike video cards, when I say gig, I mean the speed of the card (actually 1 Gb/s, gig is just easier to say). With as many users as you are planning to support, you will need a large amount of bandwidth to ensure that the NIC is not a bottleneck on your server. The reason for 2 Gig NICs (two physical 1Gb/s cards) is to again provide redundancy. What ever server you choose to go with, it should come with some sort of utility that allows you to Team (called Bonding in Linux) the NICs and create 1 virtual NIC out of the 2 physical NICs. If one fails, all traffic is routed through the second NIC in the team.


3a) What are JBODS?
-Just a Bunch of Disks. It's a way of adding adding additional external storage to a single server.

3b) You say to reach my 5tb target, I need multiple JBODS? well why cant I have a 3 Dell PowerEdge 2850 rack server for the webspace which supports 1.8TB internally and then if more space is needed, I could attach more hard drives externally to the servers?
-Each JBOD has 14 HD bays. They don't make SCSI drives big enough to get 5TB of storage out of just 14 HDs.

3c) What is a SAN? and how would it provide data security?
-Storage Area Network. A SAN provides security because it assigns storage to a particular Host Bus Adapter (HBA, like a NIC only for SANs) MAC address and will only allow access to that particular "drive" from the assigned HBA. All this is done by using the really simple web interface. It really is a simple as clicking create drive, selecting the size, and assigning the new drive to a particular HBA. After that, Windows will see it just like it is a regular HD.

3d) What are HBAs and NAS?
-Host Bus Adaptors and Network Attached Storage. A NAS is different than a SAN in that it uses regular NICs where the SAN requires HBAs and a SAN switch. The NAS (as I understand it) assigns storage by IP so any server that connects to that IP will be able to access the "drive"

4a) What is a DR strategy?
-Disaster Recovery. This means that if something goes wrong, this is your plan to get everything back to 100%. For instance, a fire in your Data Center (DC) destroys all your servers and localy stored information. What is your plan to bring all your servers back up in the same condition as before the failure.

4b) What is a DC?
-Data Center

4c) What is a tape libary?
-A tape library is an automated backup solution. Tape capacity depends on the type of tape you choose. I believe the the max un-compressed capacity of the top of the line SDLT tape is about 300GB. I'm not going to do the math on this but you can see it is going to take more than one tape to backup all that data. You have two choices, either you sit there and swap out the tapes as they get full or you have a tape library with a robotic arm do it for you. Same thing applies during a restore.


Sorry everyone for another huge post.
 
I should also say that after I made the original post, I remembered that you can get a clustered JBOD solution that should allow you to connect multiple servers to one set of JBODS to provide redundancy at the server level (if one server fails, traffic can be redirected to the good server and that server will have access to the same data as the one that failed).
 
Thanks black, but I still need help. Read my post above yours, there are questions about SAN I need to know. Can anyone help?

and also, what security does a typical server have? (Both hardware, software and physical)

Thanks for any help, the assignment is due in next week!
 
Someone please... My assignment is due in a couple of days time, and ive almost done. I just need to talk about software and hardware security for the servers.

Can anyone help?
 
Sorry for the delay Deadman

The link you posted is to a SAN component called a shelf. The shelves hold the Fibre Channel Hard Drives. The shelves themselves connect to a Fibre Channel Controller (this one also seems capable of connecting directly to a server as well). The Fibre Channel controller connects to the Fibre Channel switch. The data path would look something like this

Server (with HBA) - Fibre Channel Switch - Fibre Channel Controller - Shelf - HD

To ensure the maximum level of redundancy, you will need two HBA's for each server, two Fibre Channel switches and shelves bought in multiples of 2.

Here is a link to something similar from HP
http://h18006.www1.hp.com/products/storageworks/enterprise/index.html

This is a full SAN cabinet where the controllers and shelves are installed. The Fibre Channel switches would be located in an adjacent rack. Each server would have two Fibre runs (one from each HBA) running to the switches (one run to each switch). The switches would then use their uplink port to connect to the Fibre Channel Controllers and the controllers would each connect to all shelves in your SAN cabinet.

As far as backups go...
You shouldn't need 300GB hard drives in your servers as that will only hold your applications and OS. All your data will be stored in the SAN so you can save money by going with smaller HDs. Two 36GB drives per server (in RAID 1) should be enough to install an OS and a few apps.

To back up all the data on the SAN, you could go with the Tape Library. To implement this, you would need 1 server to control the library and to orchestrate the backup process. This server would then communicate with each server in your environment and copy all the data you have selected to the tape library on a schedule you specify.

This sort of setup, if your business can afford it, would provide the most scalable and redundant environment for your project.

It's always good to provide options when making a proposal like this. The above would cost more initially but would be lower cost in the long run.

Another solution you could recommend would be something like this.
http://h18004.www1.hp.com/solutions/enterprise/highavailability/dl380/index.html
This is an off the shelf clustered solution for high availability nodes. The only problem with this is you can not scale each cluster up past 4 nodes. If you need more than that in the future, you would be stuck with investing in all new hardware. On the plus side, this is much easier to implement and a lot easier on the pocket books of a smaller business.

You could use the same Tape Backup solution on this setup as with the SAN setup.
 
Thanks for your reply black but it is a little late. I have 2 days left to complete this and I still have alot more to do.I have already written about the SAN, the servers etc. Going back now and having to learn all this new information you have given me as well as selecting the right hardware would take time that I dont have.

The assignment did say to be brief and it did say that I wont lose marks for not getting everything 100% correct. (As long as ive made a good attempt, I am fine)

As far as backups go...
You shouldn't need 300GB hard drives in your servers as that will only hold your applications and OS. All your data will be stored in the SAN so you can save money by going with smaller HDs. Two 36GB drives per server (in RAID 1) should be enough to install an OS and a few apps.
Click to expand...
The reason for having 2 300GB hard drives is because 100 000 customers need 50mb of storage space each. Ive done the math so I need 17 servers with a 300GB hard drive in each. The additional 300Gb hard drive is for redundancy.

I have to push on with this assignment because I just want to get it done and over with. I need to write a little bit about software and hardware security for the servers such as antivirus programs, and software/hardware firewalls.

What should I write for this? I dont think I can simply install Norton on a server. Would one of these hardware firewalls be ok for my whole LAN? http://accessories.us.dell.com/sna/productlisting.aspx?c=us&l=en&cs=555&category_id=5855&first=true
 
Last plea for help!!!!

Many have helped me alot in the last few days, why stop now? My assignment is due in a few days and I need to talk about server security (both hardware and software)

PLEASE, if you can tell me anything about server security, this would be great.
 
Yes, choose a high end switch with a built in configurable firewall (one that will allow you to confiure the rules).

Norton or Mcafee will be fine for desktop antivirus protection

If you go with Windows 2003 SP1, you will have a built in Firewall on each server. Otherwise, you can choose to install a third-party software firewall solution to compliment the firewall on the switch.

Choose 2 servers out of your exisiting environment (or pick up to low end servers for the task) and install Active Directory. This will enforce user level security on each system as well as allow you to configure domain policies for securing each server that has been joined to the domain.

You can add additional security by setting up static IPs on each system and not running DHCP in your production environment. This will prevent a malicious user from easily joining a rougue server to your production environment.

Lock down the server itself by adding a BIOS password and configuring the server to boot from HD as the first option.

Buy a lockable server rack and keep it locked to prevent hands on access to the hardware.

Add security to the door leading into your data center.

Windows 2003 is better that 2000 because it comes with nothing turned on. What this means is you, in theory, don't have a bunch of listening services (security holes) on your server that aren't being used. On top of that, MS provides utilities to harden the existing security on their platforms.

Is that enough, or do you need more?
 
Thanks alot black, that seems perfect. Just one thing... what antivirus software should the servers have?

thanks again, you have been a HUGE help!
 
Well thanks everyone, you have all been a huge help, especially you black!

My assignment is complete and I will be handing it in tomorrow.,so thanks everyone!
 
You must log in or register to reply here.
Back
Top