Active Storage Announces 'ActiveSAN' Replacement for Xserve

The solution is to use http://www.centrify.com/ or http://www.likewise.com/ with Microsoft 2008 Server. That's true. it is a great solution.

Apple don't care about consumers and soon or later they will pay for this again like old bad days off apple that loyal consumers (design, video and audio) have supported the company and the ones that apple are screwing now...
 
Only 4 drives? Why not 8 2.5" drives? Or at least 6?

This isn't meant to be a complete storage solution. They have other offerings for that. This is meant to be a metadata controller.

According to Active Storage's website, metadata can eat up a lot of your storage space. Evidently a metadata controller can offload that responsibility and free up space and resources on your SAN.

This was from a quick search; I gladly defer to an expert in this area.
 
+1
I realize that this was not a server replacement, and as I read through Active Storage's web site, I'm still confused about what a MDC is, but it definitely is not a server. This product IS a solution for metadata control.

Sheesh. It *is* a server. It is a metadata server. In very much the same way a NetApp server/appliance is a server. It is not a generic utility server which you can deploy a variety of sever apps to, but it most certainly allocates and manages shared resources to clients.

XSan is a Apple rebranded implementation of StorNext clustered file system. In that file system, the metadata ( the "inodes" if you want to map it back to classic Unix file system concepts) is stored separate from the data blocks. There are advantages to doing then when going to have a serveral machines going to be competing for read/write/updates for the same blocks (e.g., lock contention, serializing updates, redundantly duplicating all the metadata on multiple "servers" without necessarily duplicating all of the data). The "data" storage device can just pump out data while another device pumps out the metadata about where the data sits on the med-high end storage device.

There are several cluster file systems that work in a similar way ( e.g. Lustre, pNFS, etc. ).

What ActiveStorage did here is drop the underlying MacOS X infrastructure because it wasn't really providing anything unique and critically essential. StorNext is multi-platform before Apple licensed it. XSan was easier to use and somewhat more affordable.

They had already talked of releasing a product InnerPool which pulled much of the Metadata serving out of the the MacOS X layer anyway. Likewise lots of storage vendors have appliance boxes where you can get to a web storage management interface.
 
Sheesh. It *is* a server. It is a metadata server. In very much the same way a NetApp server/appliance is a server. It is not a generic utility server which you can deploy a variety of sever apps to, but it most certainly allocates and manages shared resources to clients. ...

True, but it wasn't an Xserve replacement, which is what many people were expecting. I should have been more precise in my response.
 
Isn't most of the stuff in Mac OS X Server open source? It pops up about 1000 Linux distroes everyday (soon there will be more distroes than human on this planet), so why isn't there a Linux based to implement most of the features of Mac OS X Server out of the box?
 
This isn't meant to be a complete storage solution. They have other offerings for that. This is meant to be a metadata controller.

According to Active Storage's website, metadata can eat up a lot of your storage space. .

percentage wise it is not alot of storage media space. If it is then probably don't need XSan in the first place ( since each client accessing is $800-900).

The metadata has significantly different access patterns and traffic. The data is also following different routes. In XSan case , the metadata is traveling to-from the MDServer and clients while the "bulk" data is going between the clients and the storage device. You can use the MDServer as a client to pull the metadata out of the storage device also but that is an extra hop. If it is local to the MDS then can get rid of latency and frees up storage device to deal with fewer access patterns (head seeks for metadata don't get tangled up in those for "bulk" data. )
 
Why would they do that? That's like saying, "Well, our own operating system isn't good enough to use as a server so we're switching to Linux"

Maybe Apple should just port its server tools to Linux (and/or BSD)? They clearly are not interested (nor really competitive or successful in the server hardware market). They have an interest in there being servers in the market that serve Mac OS X clients well.
If you need to administer or simply use servers/workstations remotely, you do not care how the hardware looks, nor whether you can run the OS X GUI or even other commercial OS X apps on them. You want nice remote admin tools. Of course it is nice if you can test some server-side applications on your local Mac and then have them run exactly the same on the server but other than that, I do not really care what OS my *nix server/workstation is running.
 
True, but it wasn't an Xserve replacement, which is what many people were expecting. I should have been more precise in my response.

OK. I'm not sure why folks had that expectation though. If folks looked at Activestorages whole website they would see they mostly cater to high end storage needs of the digital media market. Not even high end generic storage; specialty storage. So if they are in the specialty appliance business why would the introduce a generic server ?

As a hook, to generate tons of ad views when it doesn't pan out .... sure I can see how that expectation could take hold.


I kind of wonder if there is a chicken-egg thing here to. If ActiveStorage (and some others) was thinking about doing this anyway before the XServe announcement then it was also a contributor to the shrinking XServe market. The more "attitudes" I read about the XServe cancellation the more I think it is a two way street. Large blocks of customer "quit" on Apple concept of selling systems (software+hardware) about as much as Apple quit on "enterprise" (i.e., large , not necessarily business ) market.
 
No, everyone thought this was going to be some kind of magic partnership with Mac OS X Server running on third party hardware.

As I and others said in the original story, it's nothing more than an Xsan Metadata Controller replacement running on Linux.

THANK YOU. I think we can all agree that this is not a production that "Fill the void left by Xserve's departure
 
If by "fill the gap" you mean fill one role, yeah. If by "fill the gap" you mean AFP, iChat, Calendaring, Mail, MySQL, NetBoot, NFS, Push Notifications, Software Update, and Apache (you know, just to name a few) then um... no. :confused:

The role, yes. Of course it won't do all of the things Mac OS X Server does. I didn't expect that, and anyone that did would be foolish. I did assume Apple would have a huge role in the hardware, but in terms of the software, it's still go Apple or go Home - and I'm not sure if that's good or bad still.

...if the ONLY thing you need is an Xsan Metadata Controller. We don't use Xsan at all, so this is useless to us (and many, many others) as an Xserve "replacement". For its target market, it's great indeed.

Pardon me for being foolish, but aside from storage and rack-mount-ability, what does / did the Xserve do that the Mini with OS X Server can't now? (- and I'm not being sarcastic. Genuinely curious.)
 
How is this a replacement when it doesn't run OS X? I wish Apple would make a statement about the direction they are going with OS X Server. We have switched everything to Macs here and depend on iChat, Address Book, & Push servers and there's no way in hell I'm switching everything back over to Windows.

Yes you be nuts to go to Windows. But Linux based servers work well. That is what this box is. It's Linux based but they provide a Mac OSX application to configure it so don't have to learn how to edit text files.
 
It would have to be FREEBSD but they don't even give a **** about making a gnome GUI back end to make it easy to install software.
If they would get their head's out of there butts FREEBSD could be a competitor.


Isn't most of the stuff in Mac OS X Server open source? It pops up about 1000 Linux distroes everyday (soon there will be more distroes than human on this planet), so why isn't there a Linux based to implement most of the features of Mac OS X Server out of the box?
 
Pardon me for being foolish, but aside from storage and rack-mount-ability, what does / did the Xserve do that the Mini with OS X Server can't now? (- and I'm not being sarcastic. Genuinely curious.)

It's like a Cessna and a Boeing 767. Both can fly. But that's about where it ends if you need to support more than 10 people with a robust server. There's the issue of pure scalability for intensive server operations. The lack of card slots for adding fiber, SAS or eithernet cards. There's no redundancy for data unless you use up a precious FW800 port. There's no power supply backup. It's just not an option for a company that needs many of the services that are available on Mac OS Server and have more people than the mini can handle.

-G
 
Seems that they are saying just that by discontinuing their only real server.

Its clear that Apple wants out of the server market. The least they could do is open source their server apps. Apple released Quicktime Streaming Server to the community as Darwin Streaming Server (runs on FreeBSD and Linux).

Final Cut Server, WebObjects, PodCast Producer etc... are great pieces of technology that have untapped potential.

They can live on if open sourced or go the way of Shake if Apple decides to EOL them.

-z3r0



Why would they do that? That's like saying, "Well, our own operating system isn't good enough to use as a server so we're switching to Linux"
 
Most Mac OS X Server customers will either transition to Linux, or pull their Mac OS X Server systems in-house to run on Mac Pros or Mac minis (which we don't support in our datacenter).

This is why the earlier sweeping generalization that Mac OS X Server is doomed isn't really based on markets, customers, or availability of suitable Apple system products. Just because rigid datacenters don't install it doesn't mean the product would be sold to significantly large number of customers. If anything, the number of pre-installed Mac OS X Server systems sold now is probably at an all time high.

The "centralized" vs. "distributed" server pendulum has swung to both sides of the extreme ( the "holy grail machine room is only suitable place" to "white boxes breeding like rabbits in office cubby locations"). The more optimized answer to the overall problems/needs is more toward the middle than either extreme.

Nobody needs a centralized "chilled air" , highly regimented datacenter to provide a server with 1-2 TB of data anymore. If Mac OS X Server addresses the "right sized" problem space it should have no problems selling in sufficient numbers for Apple to keep it going. It has got to make decent money though at a scale where Apple has some leverage. Apple isn't going to carry it as some charity case.

At some point perhaps Mac OS X Server will get subsumed into a variety of appliances ( more robust "Time Capsule" or "Airport" device. A "home" / "departmental" / "Small Business " appliance. )

What Apple does will speak louder than any "forward looking statement that isn't binding". A slightly tweaked Mac Pro (delivered when a new Mac Pro is due) that is more easily rackmountable as an option would address a large segment of folks who really only need 2-4 rack mounted servers. Will it conquer the whole datacenters? No. Will they sell a decent number of them? Probably yes if the software really adds value.
 
It was a sad day when Apple announced that they would discontinue the XServe. However, unless you use specific Apple software, they are a bit expensive. At any rate, I don't like to see any of my fellow Mac users suffer from Apple discontinuing a product line.
 
This is why the earlier sweeping generalization that Mac OS X Server is doomed isn't really based on markets, customers, or availability of suitable Apple system products. Just because rigid datacenters don't install it doesn't mean the product would be sold to significantly large number of customers. If anything, the number of pre-installed Mac OS X Server systems sold now is probably at an all time high.

The "centralized" vs. "distributed" server pendulum has swung to both sides of the extreme ( the "holy grail machine room is only suitable place" to "white boxes breeding like rabbits in office cubby locations"). The more optimized answer to the overall problems/needs is more toward the middle than either extreme.

Nobody needs a centralized "chilled air" , highly regimented datacenter to provide a server with 1-2 TB of data anymore. If Mac OS X Server addresses the "right sized" problem space it should have no problems selling in sufficient numbers for Apple to keep it going. It has got to make decent money though at a scale where Apple has some leverage. Apple isn't going to carry it as some charity case.

At some point perhaps Mac OS X Server will get subsumed into a variety of appliances ( more robust "Time Capsule" or "Airport" device. A "home" / "departmental" / "Small Business " appliance. )

What Apple does will speak louder than any "forward looking statement that isn't binding". A slightly tweaked Mac Pro (delivered when a new Mac Pro is due) that is more easily rackmountable as an option would address a large segment of folks who really only need 2-4 rack mounted servers. Will it conquer the whole datacenters? No. Will they sell a decent number of them? Probably yes if the software really adds value.


how are you going to backup your servers if they are spread around 20 offices ?
 
The solution is to use http://www.centrify.com/ or http://www.likewise.com/ with Microsoft 2008 Server. That's true. it is a great solution.

Apple don't care about consumers and soon or later they will pay for this again like old bad days off apple that loyal consumers (design, video and audio) have supported the company and the ones that apple are screwing now...

Centrify and Likewise handle only a small portion of what the Xserve does.

If anything you would need:
Microsoft 2008R2, ExtremeZI-P iSCSI into a NetApp, JAMF Casper, JAMF Casper SCCM API is great too, and Centrify.

Notice what this is? A lot of bandaids.
 
It's like a Cessna and a Boeing 767. Both can fly. But that's about where it ends if you need to support more than 10 people with a robust server. There's the issue of pure scalability for intensive server operations. The lack of card slots for adding fiber, SAS or eithernet cards. There's no redundancy for data unless you use up a precious FW800 port. There's no power supply backup. It's just not an option for a company that needs many of the services that are available on Mac OS Server and have more people than the mini can handle.

-G

if you had more then 10 people you would just get two minis :p

I'm kidding ofcourse.
 
It's like a Cessna and a Boeing 767. Both can fly. But that's about where it ends if you need to support more than 10 people with a robust server. There's the issue of pure scalability for intensive server operations. The lack of card slots for adding fiber, SAS or eithernet cards. There's no redundancy for data unless you use up a precious FW800 port. There's no power supply backup. It's just not an option for a company that needs many of the services that are available on Mac OS Server and have more people than the mini can handle.

-G

Hm... Now I've learned. Interesting. Xserve was worth more than I ever figured. I can see the arguments now.

I've never dabbled in the Xserve world, or Xsan at all.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Back
Top