“When eWEEK Labs first reviewed Sun Microsystems’ Solaris 10 early last year, we were impressed by the new facilities the operating system offered for better serving up applications and making the most of the SPARC and x86 hardware on which it runs. With this summer’s Solaris 10 update, labeled 6/06, Sun has significantly improved on its already excellent operating system with the addition of the much-heralded Zettabyte File System.”
to me it seems with solaris 10 sun has really made a lot of good decisions. the file system is good. the networking is good. usablility is good. and over all proformance is good. personaly i think the evolution of this operating system is really going quite smoothly. the only thing i wish they had done (and i know this will cause a little contiversy, but thats not my intent in posting it) would either be to have bought out troll tech to use QT and make a simple and farmiliar windows like interface insteat of using GTK. though useing GTK and gnome was good due to its vast amount of GTK based software it alowed to be run easily, on a level of what is probibly better technology i would have gone with QT myself… just my opinion
Edited 2006-10-24 22:07
Unfortunately, that is commonsense, something that is completely alien to Sun. If they had commonsense, they would have stuck with Solaris x86 6 years ago rather than killing it off in favour of going down the Linux track.
why to switch to an inferior technology? just because it sounds cool?
The combination of:
* performance – eg project fire-engine, ZFS
* flexibility – eg containers, brandz
* innovation – eg DTRACE
* stable consistent platform – Sun has a long history in UNIX
* improving hardware support – from accelerated nvidia to broadcom ethernet, from sound on my Dell laptop to “just works” 1680×1050 display, even USB keys are mounted
* price – almost gratis free even for commercial use
* very sound development model – see interaction between sun employees and the public community at opensolaris.org
will make it very likely that Solaris will become the primary OS used in server rooms and on developers desktops, if not in kiosk mode.
application support is getting there (adobe flash, realplayer are good signs). hardware support is improving unlike the half-hearted attempts at earlier x86 ports.
i’m seriously impressed with Sun’s level of commitment to OS innovation, and to opening it up to the public.
performance, flexibility, innovation, stable consistent platform, improving hardware support, price, very sound development model
Exactly, it’s a pity I can’t give you more points. Sun has managed to make reality out of many buzzwords that other companies only use every so often, but generally miserably fail to realize.
“For example, with ZFS, adding a new RAM chip to a system does not require partitioning or explicit allocation operations—you just add the RAM stick, and the operating system figures out how to use it.”
huh?
ZFS, Ram ??? – partitioning???
Do they actually KNOW what they’re writing about ?
yes…
with ZFS you add a disk to a pool and its recognised and its ready to be used.
no need to reformat or anything just run one command and its ready to be used.
I am well aware of the features of ZFS
– it was just a bad case of journalism comparing RAM with harddrives!
Actually it’s a valid comparison to some extent. When you add some RAM to system all applications can use it right a way. The same is with ZFS – you just add disk to a pool and all file systems in a pool can make use of it.
true, but the author wrote that you added it TO ZFS and that you didn’t need to partition it – which is a total fubar
… But if you remove a DIMM for your system, your machine remains just as stable as long as you have enough memory to boot the OS.
I doubt that you’ll get the result if you remove the root drive.
Bad journalism at its best.
– Gilboa
I think he’s trying to compare adding a hard drive to adding a stick of ram, but just made a mistake that should have been caught.
Probably should read like this
Based on our experience so far, ZFS was worth the wait. ZFS is designed to make storage management on Solaris more like memory management. For example (REMOVED) adding a new RAM chip to a system does not require partitioning or explicit allocation operations—you just add the RAM stick, and the operating system figures out how to use it.
With ZFS, administrators create storage pools out of physical disks and then create file systems that draw storage from these pools. There’s no need to preallocate sizes for ZFSes—the file systems draw from the pools as needed. We could, however, assign quotas within our ZFS pools to ensure that particular file systems had enough storage to meet their needs.
I’ve heard a lot of hype about ZFS, read the docs, watched the demos, etc. This is not intended as a flame, but I am an AIX consultant and I can’t, for the life of me, figure out what ZFS brings to the table that you can’t do with LVM+JFS2 on AIX. It is being pitched as a technological advancement light-years ahead of any other UNIX vendor and I just don’t see it.
Can someone enlighten me as to why ZFS is so much better than anything else that’s out there?
For one, it’s just ZFS instead of LVM and a file system. It’s integrated, it’s easier.
Second, it’s got RaidZ. It’s supposedly better than e.g. Raid5. I don’t know much about it though.
Third, is the cryptographically strong filesystem integrity checking. Raid can recover from a drive dying totally (well, not RAID0), but if the drive is flaky and gives false data, ZFS will detect that _and_ attempt to fix it on the fly.
Fourth, ZFS features cheap and fast filesystem snapshots. I don’t know about JFS2 though.
Please correct me if I’m wrong.
> For one, it’s just ZFS instead of LVM and a file system. It’s integrated, it’s easier.
A good point, integration is good. However, on AIX the LVM and FS components are very well integrated from a sysadmin point of view. True, they are separate components, but expanding a filesystem is one operation, creating a new filesystem is one operation, taking a snapshot is one operation. The commands used integrate the LVM and FS steps to do each of those tasks.
> Second, it’s got RaidZ. It’s supposedly better than e.g. Raid5. I don’t know much about it though.
I probably need to read more about RaidZ, but my cursory understanding is that it is basically software RAID, albeit integrated with the FS/LVM layer. I still think I would prefer hardware-based RAID.
> Third, is the cryptographically strong filesystem integrity checking. Raid can recover from a drive dying totally (well, not RAID0), but if the drive is flaky and gives false data, ZFS will detect that _and_ attempt to fix it on the fly.
You may have a good point there. I’m not aware of that capability in JFS.
> Fourth, ZFS features cheap and fast filesystem snapshots. I don’t know about JFS2 though.
JFS2 can do snapshots quickly and without much disk space. I think it’s pretty much implemented the same way UFS snapshots are on FreeBSD. Could be wrong about that, though.
#3, the integrity checking, is very important as it helps detect “silent” corruption happening on a disk drive, say from a flaky controller, and gives the system opportunities for correction, but minimally it provides notification. Effectively, ZFS is running and tracking checksums at the block level.
The more front facing benefit from an admin point of view is simply that the ZFS command line is REALLY simple and easy to use. It’s well documented, and almost “intuitive”.
To be fair, this is “syntactic sugar”, and the IBM admin app (I forget the name not have touched AIX in over 10 years) is certainly easy to use from a “green screen”, Curses style application, but from the command line it’s not as clean and clear as the ZFS command line. And compared to the de facto VERITAS system that ZFS will most likely place in large scale use, the interface is simply night and day.
I’m really hoping that Apple is looking at ZFS VERY seriously, because I have great confidence that they can make an admin tool and leverage ZFS capabilities to push it to the next level of making it a consumer friendly file system. The “adding a stick of RAM” analogy is very apt to how disk pools are managed, and I think there’s great value for consumers to be able to convert a 100G volume in to a contiguous 200G volume with simply slapping in a new hard drive.
> To be fair, this is “syntactic sugar”, and the IBM admin app (I forget the name not have touched AIX in over 10 years) is certainly easy to use from a “green screen”, Curses style application, but from the command line it’s not as clean and clear as the ZFS command line.
SMIT (Systems Management Interface Tool, I think)
> …I think there’s great value for consumers to be able to convert a 100G volume in to a contiguous 200G volume with simply slapping in a new hard drive.
True, but there’s also a lot of danger in it as well. It might give many home users a false sense of security…”I have two drives so it’s okay if one fails”
Chicklin,
I know both LVM/JFS(2) and ZFS, for me, the most significant features of ZFS are:
1) End-to-End Data Integrity.
http://blogs.sun.com/bonwick/entry/zfs_end_to_end_data
2) Lightweight and near-zero-administration filesystems.
On the other hand, LVM/JFS2 has this most significant advantage: Predictable performance.
This is first due to non-COW (most other FS) v.s. COW (copy-on-write, like zfs).
Also, when you add a disk to a VG, you still have the control over how to use the disk for differnt LV/FS, whereas adding a disk to a zfs pool is really like adding a memory stick to the system.
Unfortunately, disk is not the same as memory, yet.
The difference in ‘seek latency’ is day and night.
There’re many other differences, having different significance for different people.
It’s all about “architectural choices and trade-offs”.
The Oracle ASM is capable of rearranging used blocks and it is possible to reduce storage pool size. ASM had something like hotspot movement, don’t know, if it works. At the present moment zfs can only increase the size.
Most probably Sun will include the ability to shrink a zfs partition. I am waiting for it – freedom. Maybe they will be able to address some of taos’s concernes about performance unpredictability.
zfs & hardware raid. Maybe Sun figures, that if you need a hardware raid, then most of the times you could also buy a storage box with a hw raid.
zfs is imho at the present moment not with all the features planned for it. Maybe we should wait.
OsX – I really would like to have some zfs counterweight to the linux hype.