So on the one hand I'm saying MRAM could be great for desktops, and on the other hand you're saying "we want a server in the house".
Lokiomega is right, things like video (especially HD) are extremely bandwidth intensive, especially if you are viewing one movie, while writing another.
Modern desktops are more than capable of handling desktop and video-server tasks.
As for app binaries, Oracle recommends you have an 18GB filesystem (disk or partition) that is private to each RAC cluster member and the Oracle binaries and startup files are located in there. The Oracle RAC installation process then copies parts of the Oracle kit that need to be cluster common into a common area which I think exists inside of ASM.
As someone else pointed out, there's a vast difference between HD meant for media file storage and HDs meant for OS and application storage. The latter do not need to be nearly as big as the former, but for them speed is much more critical. If MRAM is as fast as today's DDR2, then it will be several orders of magnitude faster than hard drives. That performance difference (as well as the reliability and power improvements) makes your dollar/gigabyte comparisons completely irrelevant. People who need it (businesses) WILL buy it
But h/w manufactures already use (lots of) DRAM as cache to speed up disk access.
(I'm thinking here big NAS/SAN cache, not OS-managed cache.)
A very small 10RPM drive for the OS, programs, and a much larger (but probably slower) drive for storing all my files.
That does not sound very smart.
Maybe my thinking is clouded by experience with databases, but data should be on the fastest drives, since you need it quickly, but program load is infrequent, so if it's on a slow disk, no big deal.
Actually it wasn't DEC Rdb they got it from, it was the DLM (Distributed Lock Manager) code from Tru64 Alpha's TruCluster that some twit in Compaq sold them. Sometime round 1998 if I remember right.
Ah, you're right, I forgot all about that.
But wasn't TruCluster a reimplimentation of the VAXcluster layered product?
My worst fear was that they would take the DLM, built their own platform neutral CFS implementation, and Tru64's DLM + CFS would become irrelevant.
When a product like VAXclusters and DLM have been around for 20+ years, you've got to figure that *someone* is eventually going to reimplement it.
I've always been surprised that IBM & Sun & MSFT never reimplemented DLM to make VAX-like clusters. MSFT certainly has never had any qualms about "borrowing" other companies' ideas. Patent protection and NIH Syndrome are the only reasons I can think of why it took this long...
In fact ASM pretty much eliminates the need for any Vendor cluster filesystem. What other products are there that use CFS + DLM in the way that RAC does? I can't think of any, which also makes Veritas one of the big losers in this story.
I, as a long-time OpenVMS user, can't imagine not having a cluster-aware FS. How do database apps running on RAC nodes share scripts, flat files, etc without a DLM and cluster-aware FS?
How much more integrated can it get, shy of running in kernel space? There's already an Oracle-specific file system which is used in RAC installs. It's the only filesystem of its kind which is in the main Linux kernel.
You mentioned it: run it kernel space. Or rely on system calls only available to Linux.
Sure, Oracle wrote OCFS* (from knowledge they gained by purchasing DEC Rdb), but that does not mean that OCFS* can only be used by Oracle. It's a filesystem "just" like any other FS, in that/home,/var, etc could be on it, just like SGI's xfs or IBM's jfs.
If all these politicians got paid off, why didn't they rescue him?
Back when Enron collapsed, The Left was saying that since W was Lay's big buddy, W do something sneaky to protect Lay. Lay even called up the Secretary of the Treasury, asking for regulatory help, to pull off the SEC "dogs". No dice.
Lay had 5 years to cut a deal with the government, and never took it.
(and some technical ones -- massive changes to support your local language may be very important to you, but the original developer may not accept them).
You'd think that the main developer would welcome i18n patches.
You get alarms coming from inside the unit and you have no idea what the fuck is going on until you reboot and drop into the BIOS.
That's a pathetic enterprise, then.
A Real Enterprise doesn't use Dell Crap. It uses Tier-1 h/w and proprietary operating systems that the vendor has certified all works together. And if it doesn't work right, your EVP calls their EVP, and Action Is Taken, even if that means the Senior OS Engineer lives at your DC until it's fixed.
I've found these to be engineered to a much higher tolerances and you can always connect a TTY to it or SSH in and figure out what's going on with it.
That's sooo 1980s. DEC (then Compaq then HP) has let you "telnet" into storage controllers since 1985. Their original CI controllers were really custom-configured PDP-11s running an RTOS, and all of it's storage controllers (not cards) since then have let you plug Ethernet or a serial VT into them.
I'm sure that IBM mainframe controllers let you do the same thing, but have no direct knowledge.
This isn't about life. It's about a professional decision as someone working for a company. Before deploying half-assed solutions, it's usually better to deploy nothing at all.
That's a crock.
Every non-trivial decision needs some form of cost-benefit justification.
Sure, U320 SCSI controllers with a GB of batter-backed RAM are fast and durable, as are 15K RPM 300GB Cheetahs, 4Gbps fiber and Hitachi 9980 SANs with 128GB of cache.
However, not all data deserves to be stored in an Hitachi 9980V. And not everyone can afford them. Some organizations can only afford FAKERAID.
There's no "middle ground", there's cost-benefit analysis.
I.e., is it worth my time to spend $50, $100, $200, $500, etc, and an hour a week to mirror a pr0n collection? Some people would say $50 and 5 minutes, and others would say $500 and 6 hours a week. And some would say, "Chunk it. If the disk dies, I'll just download it all again."
Slashdot Mirror
Lokiomega is right, things like video (especially HD) are extremely bandwidth intensive, especially if you are viewing one movie, while writing another.
Modern desktops are more than capable of handling desktop and video-server tasks.
So, all we need is
- one OS: Windows
- one Office package: MSO
- one financial package: MS Money
- one development app: MSVC
MSFT must love you.If so, it's a lame one.
What a fscking hack!!!!
I am SO going to miss OpenVMS...
But h/w manufactures already use (lots of) DRAM as cache to speed up disk access.
(I'm thinking here big NAS/SAN cache, not OS-managed cache.)
That does not sound very smart.
Maybe my thinking is clouded by experience with databases, but data should be on the fastest drives, since you need it quickly, but program load is infrequent, so if it's on a slow disk, no big deal.
Ah, you're right, I forgot all about that.
But wasn't TruCluster a reimplimentation of the VAXcluster layered product?
My worst fear was that they would take the DLM, built their own platform neutral CFS implementation, and Tru64's DLM + CFS would become irrelevant.
When a product like VAXclusters and DLM have been around for 20+ years, you've got to figure that *someone* is eventually going to reimplement it.
I've always been surprised that IBM & Sun & MSFT never reimplemented DLM to make VAX-like clusters. MSFT certainly has never had any qualms about "borrowing" other companies' ideas. Patent protection and NIH Syndrome are the only reasons I can think of why it took this long...
In fact ASM pretty much eliminates the need for any Vendor cluster filesystem. What other products are there that use CFS + DLM in the way that RAC does? I can't think of any, which also makes Veritas one of the big losers in this story.
I, as a long-time OpenVMS user, can't imagine not having a cluster-aware FS. How do database apps running on RAC nodes share scripts, flat files, etc without a DLM and cluster-aware FS?
How much more integrated can it get, shy of running in kernel space? There's already an Oracle-specific file system which is used in RAC installs. It's the only filesystem of its kind which is in the main Linux kernel.
/home, /var, etc could be on it, just like SGI's xfs or IBM's jfs.
You mentioned it: run it kernel space. Or rely on system calls only available to Linux.
Sure, Oracle wrote OCFS* (from knowledge they gained by purchasing DEC Rdb), but that does not mean that OCFS* can only be used by Oracle. It's a filesystem "just" like any other FS, in that
Show us the server where Oracle 10.2 debs are, and I'll put instantly put it in my sources.list file. (No, Oracle XE does not count.)
What a fscking disaster that would be. As bad a MSFT integreating IE into Windows.
Exactly how are Ellison's statements libelous?
Because not every data storage need requires a big RDBMS.
Apparently not obscure enough...
Rubber-hose decryption works well, too.
If all these politicians got paid off, why didn't they rescue him?
Back when Enron collapsed, The Left was saying that since W was Lay's big buddy, W do something sneaky to protect Lay. Lay even called up the Secretary of the Treasury, asking for regulatory help, to pull off the SEC "dogs". No dice.
Lay had 5 years to cut a deal with the government, and never took it.
Never, ever trust a TV accent. Especially a cartoon accent.
Aren't you smart enough to know that?
You'd think that the main developer would welcome i18n patches.
What's a jerb?
Sure, an adapter card.
SCSI disk controllers let your RAID-5 sets stripe across channels to get extra bandwidth.
Maybe this is just a DEC feature, though.
Yes and no.
Yes, because the mathematics is trivial.
No, because, as you said, in order to do the calculations, all N disks must be read.
That's why, though, Big Controller Cache makes RAID-5 speed just about on par with RAID-0.
That's a pathetic enterprise, then.
A Real Enterprise doesn't use Dell Crap. It uses Tier-1 h/w and proprietary operating systems that the vendor has certified all works together. And if it doesn't work right, your EVP calls their EVP, and Action Is Taken, even if that means the Senior OS Engineer lives at your DC until it's fixed.
I've found these to be engineered to a much higher tolerances and you can always connect a TTY to it or SSH in and figure out what's going on with it.
That's sooo 1980s. DEC (then Compaq then HP) has let you "telnet" into storage controllers since 1985. Their original CI controllers were really custom-configured PDP-11s running an RTOS, and all of it's storage controllers (not cards) since then have let you plug Ethernet or a serial VT into them.
I'm sure that IBM mainframe controllers let you do the same thing, but have no direct knowledge.
That's a crock.
Every non-trivial decision needs some form of cost-benefit justification.
Sure, U320 SCSI controllers with a GB of batter-backed RAM are fast and durable, as are 15K RPM 300GB Cheetahs, 4Gbps fiber and Hitachi 9980 SANs with 128GB of cache.
However, not all data deserves to be stored in an Hitachi 9980V. And not everyone can afford them. Some organizations can only afford FAKERAID.
And some managers are stupid, cheap bastards.
There's no "middle ground", there's cost-benefit analysis.
I.e., is it worth my time to spend $50, $100, $200, $500, etc, and an hour a week to mirror a pr0n collection? Some people would say $50 and 5 minutes, and others would say $500 and 6 hours a week. And some would say, "Chunk it. If the disk dies, I'll just download it all again."
And how is what W.U. is doing any worse than the racist hysteria raised by Democrats regarding the Dubai ports deal?