08-10-2009 08:41 AM
Solved! Go to Solution.
08-10-2009 09:01 AM
Not necessarily. The iLO port allows for remote console access. You should still be able to access the console, and have all console functions, remotely. You just have to get it set up correctly. This is an ABSOLUTE MUST! You MUST have console connectivity in case something happens and the server is not bootable.
>>Obviously restores/backups would have to be done by this company....
Why? You still have network access. You can still back up over the network.
If you were doing make_tape_recovery backups, you would want to make sure someone regularly changed the tape.
>>...if you lost your internet you would be down
This is true. You should look into redundancy at your site for this as well.
>>Do you know how HP support would be affected?
It really shouldn't be. YOu would just want to make sure you have the correct equipment address for your hardware support.
08-10-2009 09:38 AM
our datacenter's are all offsite, the shortest 30min's without any regular staff onsite.
As mentioned before, remote console is a MUST have.
The only negative point is Network connection down = everything down, this will be your biggest single point of failure.
>>>>Do you know how HP support would be affected?
>>It really shouldn't be. You would just want to make sure you have the correct equipment address for your hardware support.
Check your support contracts, there are some which requires a registered location and are only valid for this.
08-10-2009 09:39 AM
Move your hardware to another company because you do not have redundant power.
I'm sorry I don't get this....what about putting redundant power into your site. Or what about a UPS, since most outages are short term. The really big outages constitute disaster, hence activate your DR plan. You can not predict a major power outages, why it might affect their site and not yours. Or maybe putting in a generator for backup power at your site is an option. Think how much they will be paying for that site - cause it's not free.
I'd be asking myself, first they're looking at moving the hardware - when do they plan to move the Admin position to that company too?! Sounds to me like they are toying around in their little 'bean counter' brains and the buzzword they are adding up to is 'outsource'.
Sorry, Lisa - maybe I'm reading something between the lines that isn't there...but...something just smells fishy to me.
But me thinks....something is awry here!
08-10-2009 10:09 AM
08-10-2009 10:20 AM
I spent a lot of my last work issue supporting systems all over the world.
If iLo is networked, and the firmware is kept current, the location of the system is of very little concern.
I supported customer owned systems all over the world and did not rack up a lot of frequent flier miles doing it.
Every once in a while a problem is going to come up that requires an eyeball visit and someone will have to actually look at the system.
HP support is paid for by CPU. The location of said CPU is of little concern to them, so long as the bill is paid.
I think you are fine so long as you have remote access to iLo.
Owner of ISN Corporation
08-10-2009 10:40 AM
I've never heard of the issue regarding "binaries" noted above (on any HP / DP combination). There are bandwith issued however, as with any transmission over a network.
If you can relocate your server, you certainly should be able to relocate the tape libraries as well. Then you wouldn't need to do backups over the network. (And you wouldn't care what *they* used).
If they are proposing using BackupExec, I'd ask them "What do you want to do with existing backups?"
BTW, I'd ask your director what he wants to do with existing backups, as they won't be useful w/ BackupExec.
08-10-2009 10:41 AM
I also quite do not understand why a generator purchase could not be approved if this server is so important that it can not be down and has to have redundant power. I think your bean counters, this may even be one of the higher level IT managers/directors, who manage by numbers rather than knowledge, have the wrong numbers. If you are talking about a single RX6600 machine, how big of a generator would you be talking about ? I can almost guarantee you, I can walk into a pep-boys and get one that can power up such a server, in the order of few hundred dollars. Of course when we talk about the enterprise data center protection levels, this number may be in the level of several thousands of dollars, but again who told your management that, relocating hardware is cheap ?
If you are really concerned that the system has to be up and accessible all the time, you, not only will need redundant power, but also need fully and seamlessly redundant network connectivity. Not to forget the need for redundant air conditioning of the remote data center, if you do not want to find yourself, facing the dilemma Dilbert had this past Sunday's cartoon.
If you are all happy with what the co-lo provider is providing you until this point, then, it is up to the sysadmin's comfort level regarding the backup solution provided. In my life as a sysadmin, I have never seen a backup solution working right out of the box in an enterprise environment. Expecting BackupExec will be the first of its kind, sounds foolish to me. You will need to have at least a few weeks of steep learning curve. And then, when you need a restore, how confident are you that the co-lo host will be able to provide exactly what you need.
Remote console is a must in a remotely hosted server case. You need to connect the equivalent of MP-LAN in PA-RISC envronment on your IA64 architecture as well as a serial solution. There are certain things that you can not do from MPLAN but can do via serial console. I am sure, being a hosting service provider, these ONRAMP people have a serial solution for the cases like yours. You just need to contact and ask them.
Last but not the least, when the push comes to the shove, does your management realize you will have to schlep yourself 3-4 hours away (and back) to help diagnose problems, which are otherwise impossible to fix ? Do they consider the cost of such trips plus the loss of productivity, which you can bet will happen in the first year or so than not ?
There is a bad smell in the decision making regarding this server relocation, even though I do not know your environment. Sounds like somebody has thrown in a box of golf balls to someone higher up in the food chain, to get this relocation look like it is better than keeping the server in its current location.
UNIX because I majored in cryptology...
08-10-2009 06:47 PM
Why does anyone care about the remote site's backup tool? You can't use it -- BackupExec is completely incompatible with DataProtector. The way that a remote site works is that someone rents you the hardware and it is the sysadmin's responsibility to install HP-UX and restore all the data. That is something that you definitely do not want any 3rd party handling. You also need to specify exactly what the networking will look like and what type of disks (SCSI, fibre, NAS) and for SANs, how to configure the arrays and switches.
For iLO, you need a special network (just like in you current data center, right?) which isolates the unsecured console access port(s). You can do this two ways:
1. Use a terminal concentrator such as Cyclades, Digi, Raritan, Avocent, etc which connect RS-232 serial consoles into a single box that has ssh and authentication built in. This box gets an isolated subnet address with access severely limited by routers.
2. Or you simply connect your iLO port to the isolated network mentioned in 1.
This is basically what a disaster recovery (DR) site looks like. It is remote so that a big problem (such as loss of all power) can be mitigated by moving the data and network access to the remote site. However, it sounds like no one has ever been trained in this specialized technology so I would look at some good DR books as well as interview DR site providers such as HP and SunGuard. Their presentations will be real eye-openers, especially for management.
08-11-2009 05:21 AM
1) You said:
>>they have redundant power companies,
>>...our director sent the networking manager to Austin a couple of weeks ago...
If you are talking about Austin, TX then the point the company makes about redundant power companies is irrelevant. In TX **ALL** power is transmitted over the same power lines, which are maintained by Oncor Electric. It doesn't matter if you use TXU, Reliant, or any other company, **ALL** power flows over the same lines. If a main line that feeds the lines that feed this data center are cut, then you are out of luck.
The one thing that **MIGHT** help here is if the company gets power from multiple separate power grids in Austin.
2) You said:
Again, this could be irrelevant as all wiring going into the data center will likely be provided by the LEC (Local Exchange Carrier). I would ask **specifically** what types of Internet circuits they have and how they are routed into and out of the building. If they have multiple entry points for cables, then this may be OK, but your biggest risk is someone with a backhoe cutting a big fibre/cable bundle somewhere outside the building. If everything is in one bundle, then you are out of luck.
08-11-2009 05:32 AM
Backup transition is a key issue that has come up.
If you intend to change backup providers, you need to maintain both licenses for a long while.
The last place I worked in the US had a 7 year backup retention policy. That means you either have to convert the old backups, which means restoring them to disk and backing up with the new software.
So take that into account with the plan.
As to power, I understand multiple companies in the US can sell power. That as noted does not change the reliability of the transmission system, there only being one in most locations.
As far as grid overlap goes, the best place for that are locations like downtown chicago, where there are three overlapping grids.
Power blackouts in such areas are very infrequent, but they can happen anywhere with the right weather.
Owner of ISN Corporation
08-11-2009 05:36 AM
If I read that right, you mean that if the server(s) are located at the remote site, they'll want to use BackupExec to backup the server, whereas at the moment you use HP Data Protector.
I'd say that was like repalcing the modern XP/Vista PC on your desk with an old 386 running MS-DOS...
BackupExec is fine and dandy on small departmental windows systems, but it is just an *awful* product on UNIX. I can't beleieve any serious third party hosting provider would expect you to use such a terrible product!
Opinions are my own, not my employers etc...