SSDs - State of the Product?

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
I've used Server 2003 as a workstation many times, and still recommend it to those with MSDN subscriptions/Power Packs. There is a distinct performance gain over XP, though there are some software compatibility issues.
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
Posting this from the OCZ 250GB as a boot drive. Initial impressions are that it is almost as fast as the Intel 80GB in terms of "snappiness". Opening a fleet of tabs in FF3 didn't cause the hesitation it did when I used the RiData 32GB, but there was a moment of waiting (a third of a second?). The boot time also was about 20% longer. I would still put it head-and-shoulders above a platter-based drive in terms of responsiveness.
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
Yup, this OCZ Core V2 250GB is nice and quick, not a bad choice for those looking for a larger capacity SSD. Unfortunately, it's poor performance with small files rules it out for our file server. It will probably live in my laptop until I find a better home for it. 180MB/s read isn't bad. ;)
 

timwhit

Hairy Aussie
Joined
Jan 23, 2002
Messages
5,278
Location
Chicago, IL
The 30GB model of the Core V2 is about $75. Do you think it would perform as well as the 250GB model?
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
The 30GB model of the Core V2 is about $75. Do you think it would perform as well as the 250GB model?

I really couldn't say. I know they gain performance by striping across multiple memory chips. If the 30GB version uses the same number of lower capacity chips, than performance shouldn't be affected. If it uses a lower number of the same capacity chips, than it would suffer.
 

Mercutio

Fatwah on Western Digital
Joined
Jan 17, 2002
Messages
22,232
Location
I am omnipresent
Actually, that's what Microsoft thinks. They're the ones who put the word "Server" in the name and included server technologies in the product.


Wrong.

There is one operating system, Windows, based on the NT microkernel. Every shipping version of Windows includes all the same system binaries and all the same optional software. The differences between the versions of Windows within a generation amount to a set of license restrictions and a few small changes in the registry and in a couple .inf files. That's it.

There are a couple other operating systems branded as Windows - Windows Mobile and the Xbox operating system - which don't use the NT microkernel or a full set of Win32 APIs, but Windows, whether it's a home, professional or server product, is all the same goddamned thing.
 

Fushigi

Storage Is My Life
Joined
Jan 23, 2002
Messages
2,890
Location
Illinois, USA
No, you are wrong. You said it yourself. MS specifically enables or disables technologies depending on whether or not the product is intended for desktop or server usage. It matters not if the code is identical. How it is licensed and how it is packaged is what matters. My only fault would be saying 'included' instead of 'enabled'.

I'd dare to also say that the pricing model for Server takes it away from being considered for workstation users. If indeed the code is the same there's no reason to pay what, a $500 premium, for Server.

But, to make it official, the 2003 license says, in 2.a., that it is licensed for a Server. Workstations and any non-server devices are not mentioned.
 

Mercutio

Fatwah on Western Digital
Joined
Jan 17, 2002
Messages
22,232
Location
I am omnipresent
It was a fairly common practice for some types of workstations to ship with a "server" edition of Windows, if the hardware in question used more than two physical CPUs. In those situations, systems would be deployed with Windows 2000 Server or Server 2003 Web Edition.

2003SWE has only a very modest price premium over Windows XP; it could be purchased for about $350 vs. $120 for a copy of XP Pro.

All editions of Windows are built from the same code. Microsoft might ship a different kernel revision with Server Editions from Desktops, but fundamentally, they are all the same operating system. Literally all the "desktop" features of Vista can be turned on in Server 2008, and it only take a few small tweaks or edits to turn on "server" features like Shadow Copies, Fault Tolerant storage or Remote Desktop in versions of Windows that supposedly do not support those things. Pick a "server" feature. Take a quality moment with Google, and I assure you that someone has turned it on in "desktop" Windows.

Ultimately, all versions of Windows are the same thing. The only practical difference is the scale of hardware being used to run it.
 

Santilli

Hairy Aussie
Joined
Jan 27, 2002
Messages
5,257
2003 Web edition supports 4 processor, or at least two, plus two hyperthreaded.

Is it possible to use 2003 Webedition on a machine with two dual core processors, or quadcores?
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
Yup. Everyone I know licenses per physical processor (Microsoft, Citrix, VMWare, etc). Regardless of the number of cores/hyperthreading.
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
Other than the page file, I'm not seeing why SSD is so important for the OS. Maybe a few seconds are saved when booting, but that hardly increases productivity and profits. It is much more useful for the applications that create large temporary files, or working with numerous small files..
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
I agree that the OS itself doesn't need an SSD, but it is too much trouble separating all the bits that I would want better performance from the OS. But hey, I did stick it on the slower Intel SSD.
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
Yes, and even most applications don't need the write speed as much as the read speed. I found that only the
 

timwhit

Hairy Aussie
Joined
Jan 23, 2002
Messages
5,278
Location
Chicago, IL
Yup. Everyone I know licenses per physical processor (Microsoft, Citrix, VMWare, etc). Regardless of the number of cores/hyperthreading.

Oracle doesn't license per physical CPU. They use crazy multipliers.

Oracle said:
For the purposes of counting the number of processors which require licensing, for a Sun UltraSPARC T1 processor with 4, 6 or 8 cores at 1.0 gigahertz or 8 cores at 1.2 gigahertz for only those servers specified on the Sun Server Table which can be accessed at http://oracle.com/contracts, “n” cores shall be determined by multiplying the total number of cores by a core processor licensing factor of .25. For the purposes of counting the number of processors which require licensing for AMD and Intel multicore chips, “n” cores shall be determined by multiplying the total number of cores by a core processor licensing factor of .50. For the purposes of counting the number of processors which require licensing for all hardware platforms not otherwise specified in this section, a multicore chip with "n" cores shall be determined by multiplying "n" cores by a core processor licensing factor of .75. All cores on all multicore chips for each licensed program for each factor listed below are to be aggregated before multiplying by the appropriate core processor licensing factor and all fractions of a number are to be rounded up to the next whole number. When licensing Oracle programs with Standard Edition One or Standard Edition in the product name, a processor is counted equivalent to an occupied socket; however, in the case of multi-chip modules, each chip in the multi-chip module is counted as one occupied socket.
 

Santilli

Hairy Aussie
Joined
Jan 27, 2002
Messages
5,257
It looks like the Intel SATA SSD's are coming of age.
So, I guess that means I need to buy a Panasonic notebook that has SATA capability.

Any suggestions?
CF 52?
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
Ugh. A great reason to never use Oracle.

Ugh? Most of the companies I work with or for use Oracle or SAP for the serious work. They generate billions or tens of billions in annual sales, so they must be doing something right. MS stuff is only for the office desktop users, and then there are special purpose apps for various groups.
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
All the specialty apps that I've used have allowed either MS SQL or Oracle at the back-end. Some are even kind enough to support MySQL (an obvious choice from a cost standpoint). Considering I'm given the choice of sticking with a vendor with a known support structure and a simple (yet pricey) licencing plan, or an additional vendor with licencing agreements like those timwhit mentioned above? Not a difficult choice.
 

Handruin

Administrator
Joined
Jan 13, 2002
Messages
13,916
Location
USA
Ugh. A great reason to never use Oracle.

When applications need the power and you need a product that scales up rather than scaling out, Oracle does a good job the more you add CPU's to it. I had read a unisys white paper a while back on scaling up vs scaling out and Oracle did very well where as others seemed to drop off early (like MS SQL, Exchange, etc).
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
Agreed. But scaling up doesn't make sense financially when you can scale out. 8 4-core boxes cost much less than one 32-core box. Granted, some problems can't be distributed that way, but most can. That is why I never understood using big iron (eg. AS/400s) to run a bunch of VMs; You were paying the big box premium just so you could break it up again.

I'm sure there are situations where Oracle is the best solution, I'm just glad I don't have to support one of them.

Citrix, on the other hand, is the large thorn in my side at the moment.
 

Handruin

Administrator
Joined
Jan 13, 2002
Messages
13,916
Location
USA
Agreed. But scaling up doesn't make sense financially when you can scale out. 8 4-core boxes cost much less than one 32-core box. Granted, some problems can't be distributed that way, but most can. That is why I never understood using big iron (eg. AS/400s) to run a bunch of VMs; You were paying the big box premium just so you could break it up again.

I'm sure there are situations where Oracle is the best solution, I'm just glad I don't have to support one of them.

Citrix, on the other hand, is the large thorn in my side at the moment.

You could scale out and only look at the cost of the individual boxes, but some data centers have a high cost to mount additional iron because of space constraints (we have this to some extent even in my own lab at work). Not only are you powering more CPU's but additional drives, power supplies, etc. You also have to cool more physical area which costs more. I'm not saying scaling up fits all needs, but those are some reasons why people might go up rather than out. There is also less hardware to maintain which can be helpful sometimes.

I've not done much with Citrix...the limited exposure I had wasn't anything exciting.
 

ddrueding

Fixture
Joined
Feb 4, 2002
Messages
19,719
Location
Horsens, Denmark
AFAIK, blade servers manage a higher core/U than most big iron. And are essentially scaling out instead of up. But I agree that it (like all things in life) is not black and white. Except that cocaine/priest thing Merc was talking about...that is flat-out awesome.

Citrix basically takes an already complex thing (terminal server for windows) and adds a significant layer of crap and cost on top. Again, I'm sure there are great uses, but I don't have one of them. We have to support it because the vendor can't write proper network printing support, but only will deploy on terminal servers.
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
Presumably there is no point in defragmenting an SSD because the memory cell allocation is dynamic, but what about secure deletion. Prior to drive disposition, is a 7-pass overwrite method preferred or is it even necessary?
 

jtr1962

Storage? I am Storage!
Joined
Jan 25, 2002
Messages
4,365
Location
Flushing, New York
Presumably there is no point in defragmenting an SSD because the memory cell allocation is dynamic, but what about secure deletion. Prior to drive disposition, is a 7-pass overwrite method preferred or is it even necessary?
No. The cells in an SSD can be in one of two states (or four in the case of MLC). Once you overwrite the entire drive with zeroes, all the cells are all flipped to the "zero" state with no residual memory of the previous state. The best analogy I can think of is discharging a capacitor to zero volts. No part of the capacitor is going to remember what voltage it was prior to discharge. The only reason for multiple wipes with a magnetic disk is because the heads don't always entirely erase the previous state of the bit. However, even with magnetic disks, it takes specialized equipment to attempt to recover info. In short, when you don't want your SSDs any more just overwrite them with zeroes and resell them on eBay. Nobody, including national security agencies, will be able to recover any previous data as it just doesn't exist any more. Therefore, you can make a little extra cash selling the drives instead of disposing of them with no fear of leaking sensitive info.

On another note, why do I always read posts where people are so paranoid about disposing of hard drives? Seriously, almost nobody has info important enough that some security agency is going to attempt to recover it. A simple wipe of the drive with zeroes makes data recover difficult enough so that it isn't cost effective even for credit card fraudsters to attempt data recovery. What gives that everyone is so paranoid these days?
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
OK, I'll use a one-pass overwrite. I was thinking that the charge states may not be exactly the same after a single erasure and that some technique could be used to obtain an analog output and process the data.

I do recall reading about 3-bit MLC memory, not just the 2-bit variety . I don't know who if anyone uses it yet.
 

jtr1962

Storage? I am Storage!
Joined
Jan 25, 2002
Messages
4,365
Location
Flushing, New York
OK, I'll use a one-pass overwrite. I was thinking that the charge states may not be exactly the same after a single erasure and that some technique could be used to obtain an analog output and process the data.
The memory cells have an extremely small capacitance relative to the circuitry used to change their state. Therefore, when erasing they'll end up pretty close to whatever state represents a zero. For example, let's say for a single level cell 0 to 1V is considered zero while 2 to 3V is considered a one. When the cell is erased, it'll probably be something like 0.01 volts. As a cell gets heavily used it might end up being somewhat higher, perhaps 0.3 volts. Eventually the cell can't reliably hold a state and the controller simply reassigns the data to a spare. Regardless, the voltage of a cell is not affected by the previous voltage to enough of an extent to reliably recover data. In fact, it shouldn't be affected by the previous voltage at all. Besides that, there is no way to read the voltage of a cell directly. The SSDs controller reads it internally and uses a voltage comparator to determine if the value represents a one or zero. However, this cell voltage reading has no need to appear on any of the output pins on the chip. The only access to it would be by physically removing the chip's epoxy case. That would destroy the chip, and also alter the states of many of the cells, making the exercise pointless.

A great analogy here is to use a rechargeable battery to represent a cell. You don't know whether a cell with a voltage representing half-full was previously a full cell which was half discharged or an empty cell which was half charged.

I do recall reading about 3-bit MLC memory, not just the 2-bit variety . I don't know who if anyone uses it yet.
2-bit MLC is in use. 3-bit would require 8 cell levels instead of the 4 required by a 2-bit MLC as there are 8 potential states. That in turn requires much more precise circuitry. I'm sure it's coming, but I doubt it's here yet.
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
Now I found that reference. Of course 8 MB/sec. is horribly slow, but good enough for USB flash drives, iPods and assorted junk.
 

udaman

Wannabe Storage Freak
Joined
Sep 20, 2006
Messages
1,209
Continuing the dd 'Kingston' SSD thread ;)

1. Not known for high reliability from what I've read (kind of what Merc thinks of WD :p), Patriot announced a V3 of their 'Warp' series SSD @256GB, currently the only SSD I could find <$1k...until the Samsung shows up.

PE256GS25SSDR

(faster than the V2 series, about the same performance as Intel X25-E SLC).
Not in stock currently...but 'available', lol.

http://www.ncix.com/products/index.php?sku=35775&vpn=PE256GS25SSDR&manufacture=Patriot

2. mid-2009 or so, we'll have the 512GB Toshiba for $$$ ??? and Sandisk should have their 240GB fast (proprietary controller, eliminates random write speed problem?) SSD @$499. $249 for the 120GB model (wonder how much cache these have, given reduced capacities).

http://www.sandisk.com/Corporate/PressRoom/PressReleases/PressRelease.aspx?ID=4478

3. slower V2 Patriot 128GB is on ZZF, along with reduced price on Intel's slow MLC 160GB...only $780 now :D.

http://www.zipzoomfly.com/jsp/ProductList3.jsp

4. Anecdotal evidence on the ^^^V2 vs Seagate 7200.3 320GB doesn't seem to make it worth the high entry price.

http://forum.notebookreview.com/showthread.php?t=300292



5. Seems the old JMicron controllers are guilty of write latency problems, the new ones alleviate that issue, but how do you know which JMicron controller the SSD uses? see post by 'dhanson865' in thread below:

http://www.silentpcreview.com/forums/viewtopic.php?p=449160&sid=12c5aa8a202b326a542e255ad7bcef11
 

Santilli

Hairy Aussie
Joined
Jan 27, 2002
Messages
5,257
Udaman, you are da man. Thanks for a great, informative post, on topic...
 

Santilli

Hairy Aussie
Joined
Jan 27, 2002
Messages
5,257
Here's my boot drive, not an SSD, but not bad...
HDTune_Benchmark_MEGARAID_LD__0_MEG.jpg
 

LunarMist

I can't believe I'm a Fixture
Joined
Feb 1, 2003
Messages
17,454
Location
USA
I think he gave up on the Mtron type of SSD a while ago. I'm thinking about putting a few X25-E's on a 9650SE. Hmmm.
 
Top