I moderated in that thread, hence my AC post. But blowing off the 2 points I spent is definitely a reasonable price to pay to find a decent sysadmin, assuming it can help people believe in the offer. -- zpdixon at gmail.com
I could make a similar list of Windows flaws/annoyances.
You have been using Windows for so many years that you have come to a state where you've accepted (or don't even realize) that Windows is not flawless either. The same can be said about any OS actually. No OS is perfect.
The ZDNet blogger claims that Intel's XEON X5365 3.0 GHz quad-core CPU shipped back in April.
However I can't find it in the current Intel pricelist,
neither at any of the popular online retailers. Only a bunch of hardware websites have been able to review this processor.
It looks like its not yet available to the general public. Am I missing something ?
The best reference would be the official Solaris Hardware Compatibility List (HCL). But here is a piece of advice: the whole family of nVidia nForce chipsets is generally well supported by Solaris: any motherboard based on the nForce 4, nForce 500 (and maybe nForce 600) chipset should work flawlessly with Solaris, that includes probably more than half of the market of entry-level and mid-range motherboards. In my case I wanted a cheap, low-power, GbE-enabled fileserver capable of serving files over NFS at a throughput of 70-80% of the bandwidth of a GbE link. So I bought the most inexpensive nForce 4 mobo I found on newegg, with on-board GbE (even entry-level GbE controllers are easily capable of saturating a GbE link nowadays), and with a socket 754 (so I could use it with a low-power 25W Turion processor).
Regarding the SATA controller to use, I would recommend you either the Marvell 88SXxxxx family (such as the 88SX6081: 8-port, PCI-X, about $100), or the Silicon Image 3124 (4 ports, PCI-X, about $60), or an AHCI compatible controller (such as the built-in SATA controller found in modern Intel chipsets: ICH6, ICH7, etc, but you will need to use recent OpenSolaris builds: "Nevada B56" and up). Solaris supports SATA hotplug for these 3 families of SATA controllers.
The raid 5 "write hole" problem (I should have used the double quotes like this,
instead of around the term raid 5) is mentioned here [1] and on pages
16-17 of this presentation [2].
Double parity: ZFS implements it (raidz2).
Speed: ZFS is fast because it is typically used with local disks (high-throughput & low-latency), for example: the sequential read/write throughput on a Sun Fire X4500 server with 48 SATA disks in a 4U chassis has been measured as reaching ~2-3 GBytes/s (yes, bytes not bits).
Expandability: ZFS requires no downtime when expanding the size of a zpool.
Support: Sun offers commercial support for Solaris & ZFS.
My advice to NAS/SAN vendors: keep an eye on ZFS:)
The problem comes with hotswapping. I don't know if the drivers are up to that yet
Only 4 (out of ~12) Linux SATA drivers support hotswap: ahci, sata_nv, sata_sil and sata_sil24. Fortunately the first 3 ones support the majority of SATA chips on the market.
But I also highly doubt that OpenSolaris SATA drivers for some low price chip in a low price storage box can deal with hotswapping.
Meh, you are so wrong:-) About 3 months ago, I assembled a cheap AMD64 2.5 TB ZFS fileserver running OpenSolaris Nevada B55 for $950, that's $0.38/GB (!) and it supports hotswap (Sil3124 controller).
I switched to ZFS from my previous 1 TB Linux MD raid5 setup precisely because the ZFS feature set is far superior to what the Linux MD setup offered me (end-to-end checksumming, consistency of data guaranteed after a unexpected reboot, no "raid 5" write hole, flexibility of managing multiple fs on a single zpool, compression, etc).
Huh ? ZFS or SAN, no matter what you choose, in both cases you can decide to put the file server on, say, a GbE network, so in both cases the 1 Gbps bottleneck will be the same.
Regarding disk connectivity, a typical ZFS setup (like the one described in the article) uses local SATA 1.5G/3.0G disks (so 1.5 or 3.0 Gbps per *disk*). Therefore the max aggregate throughput can be theoretically much higher than your typical SAN with one or two 4-Gbps FC links. For a fair comparison, don't forget to make sure that the SATA or FC controllers do not hit another bottleneck (PCI-X or PCIe bus...).
I have seen some ZFS benchmarks where 48 disks in the same zpool were able to provide 20+ Gbps of sequential read/write throughput.
And what happens when the RAID controller fails and corrupts all of your drives?
That's the whole point of ZFS: you don't need (and don't want) to use ZFS with a hardware RAID controller.
Sun designed ZFS so that you just feed it with *disks*. ZFS takes care of the rest: volume management, corruption detection, software RAID, etc.
Look at the high-end Sun Fire X4500 server ("Thumper") they released a few months ago: 48 drives in 4U with *no* hw RAID raid controller, Sun designed this server as the perfect machine to run ZFS.
May I add: Fault Management Framework [1], Crossbow [2], pNFS [3], stable device driver interface (one of the biggest point driver developers complain about in Linux). Clearly the GP has no idea about the number of technological advances Sun is pushing in OpenSolaris.
Actually the most recent big country having switched the driving-side of the road is Sweden. This logistical nightmare happend on September 3rd 1967, also known as Dagen H (the H day).
Most good hard disk encryption technologies behave in way that if a single bit is flipped in an encrypted sector, then the whole decrypted sector becomes corrupted (and others sectors around this one are not affected). This sort of behavior is desired and help prevent content leak attacks.
For example, Loop-AES behaves like this in multi-key-v3 mode where CBC is used with an IV computed from a secret key, the sector number, and plaintext blocks [1..n-1] in the sector. This is also how Microsoft Bitlocker behaves because they combine CBC with the Elephant diffuser. When CBC is not used, this property can be achieved using LRW or XEX, or wide-block encryption.
Of course there is a formal proof, just ask any cryptographer or cryptanalyst. A basic sketch of it is that DRM makes use of conventional cryptography. However conventional cryptography has never been designed to prevent attacks in a threat model where the attacker has illimited physical access to the device performing the decryption operation.
GSM is the most sophisticated communications protocol that I have ever seen. I have read the standard (dispite getting a headache in 5 minutes) and it is totally locked down using encryption, session keys, etc.
I am shocked to see this statement so highly moderated !
You are obviously not qualified to comment on the GSM standard.
GSM is riddled with flaws and makes use of particularly weak ciphers
that are known to be so poorly designed that communications can be decrypted
in a few seconds
with a stantard PC.
Huh ? Do they realize these satellite and aerial photos (high-res areas are actually photographed from planes not satellites) can be freely and relatively anonymously purchased by anyone from companies such as NAVTEQ ? Blurring sensitive areas in Google Earth/Maps is not going to stop "evil" people from getting access to unedited photos...
I also bought a PS3 two weeks ago primarily to develop on the Cell processor. So far, I have written a prototype application that runs 5x faster on the PS3's Cell than on the highest end Woodcrest Xeon at 3.0 GHz. When I am not coding on it, I am also using it as a Blu-ray player. Given all this, plus the fact that the PS3 is a next-gen console (though I don't plan to buy any game), I am probably one of the few to recognize that $500 is dirty cheap for such a polyvant device !
Slashdot Mirror
I moderated in that thread, hence my AC post. But blowing off the 2 points I spent is definitely a reasonable price to pay to find a decent sysadmin, assuming it can help people believe in the offer. -- zpdixon at gmail.com
I could make a similar list of Windows flaws/annoyances.
You have been using Windows for so many years that you have come to a state where you've accepted (or don't even realize) that Windows is not flawless either. The same can be said about any OS actually. No OS is perfect.
The ZDNet blogger claims that Intel's XEON X5365 3.0 GHz quad-core CPU shipped back in April. However I can't find it in the current Intel pricelist, neither at any of the popular online retailers. Only a bunch of hardware websites have been able to review this processor. It looks like its not yet available to the general public. Am I missing something ?
Also, people interested in ZFS should read these slides: http://www.opensolaris.org/os/community/zfs/docs/z fs_last.pdf
The best reference would be the official Solaris Hardware Compatibility List (HCL). But here is a piece of advice: the whole family of nVidia nForce chipsets is generally well supported by Solaris: any motherboard based on the nForce 4, nForce 500 (and maybe nForce 600) chipset should work flawlessly with Solaris, that includes probably more than half of the market of entry-level and mid-range motherboards. In my case I wanted a cheap, low-power, GbE-enabled fileserver capable of serving files over NFS at a throughput of 70-80% of the bandwidth of a GbE link. So I bought the most inexpensive nForce 4 mobo I found on newegg, with on-board GbE (even entry-level GbE controllers are easily capable of saturating a GbE link nowadays), and with a socket 754 (so I could use it with a low-power 25W Turion processor).
Regarding the SATA controller to use, I would recommend you either the Marvell 88SXxxxx family (such as the 88SX6081: 8-port, PCI-X, about $100), or the Silicon Image 3124 (4 ports, PCI-X, about $60), or an AHCI compatible controller (such as the built-in SATA controller found in modern Intel chipsets: ICH6, ICH7, etc, but you will need to use recent OpenSolaris builds: "Nevada B56" and up). Solaris supports SATA hotplug for these 3 families of SATA controllers.
I kept the list of what I bought 3 months ago:
Coolermaster RC-330-KKN1-GP Elite 330 Mid Tower Case (Black) Retail$45 http://www.zipzoomfly.com/jsp/ProductDetail.jsp?P
ECS NFORCE4-A754 Socket 754 NVIDIA nForce4 4X ATX AMD Motherboard
$46 http://www.newegg.com/Product/Product.asp?Item=N8
AMD Turion 64 MT37 Lancaster 2.0GHz (25W)
$69 http://www.newegg.com/Product/Product.asp?Item=N8
COOLER MASTER DK8-8ID2A-0L 80mm Rifle CPU Cooler - Retail
$5 http://www.newegg.com/Product/Product.asp?Item=N8
CORSAIR ValueSelect 512MB 184-Pin DDR SDRAM DDR 400
$38 http://www.newegg.com/Product/Product.asp?Item=N8
Thermaltake W0070RUC TR2 Series 430W
$40 http://www.zipzoomfly.com/jsp/ProductDetail.jsp?P
Western Digital Caviar SE 16 WD5000AAKS 500GB
$625 (125*5) http://www.zipzoomfly.com/jsp/ProductDetail.jsp?P
Silicon Image 3124
$70 http://cooldrives.com/saii3gra4p64.html
Total: $938
If you buy this today, prices would be even lower ! I would feel jealous of you having a setup cheaper than mine :-)
The raid 5 "write hole" problem (I should have used the double quotes like this, instead of around the term raid 5) is mentioned here [1] and on pages 16-17 of this presentation [2].
[1] http://opensolaris.org/os/community/zfs/whatis/[2] http://opensolaris.org/os/community/zfs/docs/zfs_
Double parity: ZFS implements it (raidz2).
:)
Speed: ZFS is fast because it is typically used with local disks (high-throughput & low-latency), for example: the sequential read/write throughput on a Sun Fire X4500 server with 48 SATA disks in a 4U chassis has been measured as reaching ~2-3 GBytes/s (yes, bytes not bits).
Expandability: ZFS requires no downtime when expanding the size of a zpool.
Support: Sun offers commercial support for Solaris & ZFS.
My advice to NAS/SAN vendors: keep an eye on ZFS
I switched to ZFS from my previous 1 TB Linux MD raid5 setup precisely because the ZFS feature set is far superior to what the Linux MD setup offered me (end-to-end checksumming, consistency of data guaranteed after a unexpected reboot, no "raid 5" write hole, flexibility of managing multiple fs on a single zpool, compression, etc).
You may be confusing it with the Community Edition, which is the bleeding-edge unsupported version.
Huh ? ZFS or SAN, no matter what you choose, in both cases you can decide to put the file server on, say, a GbE network, so in both cases the 1 Gbps bottleneck will be the same.
Regarding disk connectivity, a typical ZFS setup (like the one described in the article) uses local SATA 1.5G/3.0G disks (so 1.5 or 3.0 Gbps per *disk*). Therefore the max aggregate throughput can be theoretically much higher than your typical SAN with one or two 4-Gbps FC links. For a fair comparison, don't forget to make sure that the SATA or FC controllers do not hit another bottleneck (PCI-X or PCIe bus...).
I have seen some ZFS benchmarks where 48 disks in the same zpool were able to provide 20+ Gbps of sequential read/write throughput.
Look at the high-end Sun Fire X4500 server ("Thumper") they released a few months ago: 48 drives in 4U with *no* hw RAID raid controller, Sun designed this server as the perfect machine to run ZFS.
May I add: Fault Management Framework [1], Crossbow [2], pNFS [3], stable device driver interface (one of the biggest point driver developers complain about in Linux). Clearly the GP has no idea about the number of technological advances Sun is pushing in OpenSolaris.
[1] http://www.opensolaris.org/os/community/fm[2] http://www.opensolaris.org/os/project/crossbow
[3] http://www.opensolaris.org/os/project/nfsv41/pnfs
Ahh that was you.
Actually the most recent big country having switched the driving-side of the road is Sweden. This logistical nightmare happend on September 3rd 1967, also known as Dagen H (the H day).
For example, Loop-AES behaves like this in multi-key-v3 mode where CBC is used with an IV computed from a secret key, the sector number, and plaintext blocks [1..n-1] in the sector. This is also how Microsoft Bitlocker behaves because they combine CBC with the Elephant diffuser. When CBC is not used, this property can be achieved using LRW or XEX, or wide-block encryption.
Of course there is a formal proof, just ask any cryptographer or cryptanalyst. A basic sketch of it is that DRM makes use of conventional cryptography. However conventional cryptography has never been designed to prevent attacks in a threat model where the attacker has illimited physical access to the device performing the decryption operation.
Revocation, obfuscation, TPM chips, hardware tricks ? Whatever, DRM is provably insecure.
Would grandma be able to do that ? Looks like not even that supposedly easy-to-use OS is really that easy after all...
[Beh... apparently slashdot ate my ligature oe as well as the single quote.] The guy wrote "h'orderves" to mean "hors d'oeuvres".
The guy wrote "horderves" to mean "hors d'uvres".
I am shocked to see this statement so highly moderated ! You are obviously not qualified to comment on the GSM standard. GSM is riddled with flaws and makes use of particularly weak ciphers that are known to be so poorly designed that communications can be decrypted in a few seconds with a stantard PC.
Don't you think that a terrorist would rather ask for images of a whole city or region, instead of specific sensitive facilities ?
Huh ? Do they realize these satellite and aerial photos (high-res areas are actually photographed from planes not satellites) can be freely and relatively anonymously purchased by anyone from companies such as NAVTEQ ? Blurring sensitive areas in Google Earth/Maps is not going to stop "evil" people from getting access to unedited photos...
No, it is a password cracker. More info in a couple of week, when I release it as open source.
I also bought a PS3 two weeks ago primarily to develop on the Cell processor. So far, I have written a prototype application that runs 5x faster on the PS3's Cell than on the highest end Woodcrest Xeon at 3.0 GHz. When I am not coding on it, I am also using it as a Blu-ray player. Given all this, plus the fact that the PS3 is a next-gen console (though I don't plan to buy any game), I am probably one of the few to recognize that $500 is dirty cheap for such a polyvant device !