NASLite Network Attached Storage

No argument there, except if you are going to store data of any value on a NAS server, the buggy freenas is not the OS to do it with. Hardware can cause plenty of problems on its own. There is no need for the OS to add to the confusion. I don't care about saving $30 with freenas and causing thousands of dollars in damage due to lost data and productivity. I bet your clients will agree and happily cover the cost of naslite and the fifteen minutes of your time necessary for installation.

If you really make a living in IT, this should be second nature to you. I know that naslite is not the NAS for every case, but it does fit nicely in more scenarios than one may think. Imagine the ease of talking your client through administration of their naslite server. No bull, only a couple of easy settings, a reboot and you have a very happy customer. Happy customers amount to job security. The alternative just isn't as appealing.

OK, I've beaten this horse to death

dimension, please read my post earlier (probalby 3 or 4 earlier). FreeNAS (except the software RAID) is based on stable OS and software. The only explanation that it can have problem is bad configuration file from their webGUI. Actually I have encountered some problems myself with their webGUI, but if the config file is created correctly (meaning going through "correct" steps in the webGUI), it is absolutely stable.

Second, reading about Naslite in the forum, you will see it has its own share of problems.

jfrantz, I am aware of openfiler and will check it out. It just sounds a little heavy weighted.

mike, the cost, power consumption, and space are all factors. I would rather use SCSI (and I have only used SCSI so far in my production stuff), but it would require 14amps, $35K, and 20U if I want to use RAID-1. XServe RAID with 500GB drives will cost $30k, 9U, and probably 9amp. If I can use cheap 1U servers with 3x750GB HD's, it will cost $10K, 5amp, and 10U.

By using iSCSI, I could use drives from different servers to pair up RAID-1, so if a server goes down completely, the data is still fine.

FreeBSD is a stable OS when properly configured.

Freenas is not a stable OS even if properly configured since it suffers from poor resource management as enforced by the GUI. Why is that so difficult for you to grasp?

Freenas is based on m0n0wall, a great piece of software that I myself use. If freenas is half as good of a NAS as m0n0wall is a firewall, then we wouldn't be having this discussion. I am clear on the fact that both are based on FreeBSD. I'm also clear on the fact that freenas is not FreeBSD, so don't mix the two.

John, I realise I may be diverging, but I feel that reliance on SCSI these days is difficult to justify. The cost of SCSI drives is just too high (7x on last check). The cost of SCSI drives has not come down as much as ATA or SATA purely because large businesses and IT Managers buy from companies such as IBM, HP and Dell who push SCSI, fibre channel etc at higher margins on the premise that it is more reliable and better performing.

This was closer to the truth, as far as performance goes, up to 5-6 years ago due to the absence of options in (S)ATA RAID. Reliability in my opinion is not that much different.

These days with high performance SATA hardware RAID controllers and cheap and reliable SATA drives, I believe that in most cases, where IO performance is not the critical success factor, building a multi TB array on SCSI is just throwing money away or rather into the coffers of manufacturers.

dimension,

I don't believe the freenas developer write any code other than packaging freebsd and other open source packages and the webGUI. I may be wrong though.

On the other hand, freenas has only been going on for a few months. It is quite forseeable it will be stable (even by your definition) in another six months. My project won't go live for at least four to five months so it could be a real alternative to me.

ALucas,

Yes you are diverging SCSI is still the king for DB performance....and for most web servers, that's the only thing matters. For storage, (please see my older posts) I am looking at either NAS with naslite or freenas, SAN with freenas, or direct-attached SATA RAID. SCSI is an option but you are right, the cost is too high so it is not seriously considered.

OK, this thread has gone wild enough. Going back to the topic, I would like to see naslite to "borrow" some features from freenas in the following areas:

1. One ISO for all purposes, including CD/floppy, installing to USB, and installing to HD. Installing to HD is very important feature.
2. Some kind of real free trial. This is really a suggestion for SE to broaden their customer base and have higher sales.
3. If SE implement iSCSI, I don't even mind paying higher price for a higher level product (call it Sanlite?)

Naslite is a terrific product, but from the forum traffic, I can tell they need improvment on sales. I want the company to exist and prosper so 1. it is an alternative if I found freenas not satisfactory, and 2. we get regular updates and support.

Hello John,

Your posts have generated some interesting discussions. Usually, both Ralph and I prefer to sit those out, but your last post above warrants a response.

Firstly, a single ISO for all scenarios is a very bad idea especially for people of low proficieny. If the point is to simply get all available versions, the v1 Suite filled that void for NASLite v1. When all NASLite-2 versions are out, we'll consider a v2 Suite.

The floppy versions are more than sufficient to illustrate what NASlite is all about. The points that the guys made earlier are pretty valid in this case. Ralph had a chuckle on that one. He asked "Where do I get a free eval of OSX? or XP Media Edition for that matter?". You get the point. NASLite is an OS and not an app. If you like the floppy, you'll probably like the full release.

SANLite was Ralph's idea of about an year ago. That's what led to the remote storage option in NASLite-2. SANLite (great name - don't you think ) is in the queue, will likely support both iSCSI and AoE and may be available later this year.

In closing, don't let the forum traffic fool you. There are roughly 50,000 floppy downloads from our website this year alone. NASlite software has been distributed with both magazine publication CDs as well as a number of branded 3rd party products. Companies that have purchased and are using NASLite range from offshore drilling riggs and military installations to photo labs, law offices and shools.

The Only reason to look at SCSI, FC-AL, or SAS is the superior I/O per second of the drives. Most in that class are tuned for that rather than raw data transfer. Also the FC-AL and SAS drives have redundant channels to and from the drives.

That all said I still don't get why you are looking to configure your array in a RAID1. You could get a pair of 3Ware 9500-12 cards and hang 8-750GB drives off of each. This would give you 10TB with parity and a hot spare that would rebuild in the background if you had a drive fail. This means that you could get the storage you need with only 16-750GB drives in a RAID5. For RAID6 it would be another 2 drives for a total of 18-750GB drives with a hot spare still.

If you go with a RAID1 you need 28-750GB drives. That is 12 more drives. Throw in another card or 3 to get the extra SATA ports and the heat they all produce from the extra energy they eat not to mention the cost and you have quite a bit more expense.

The more drives and servers, the more of a chance of failure. That is just a simple fact.

Mike

I don’t know why I bother. You seem pretty determined not to get it, so this is the last thing I am going to say on the topic.

Yes, you are wrong. If you really want to understand what the author did or did not do, take a look at this document by the author: http://www.freenas.org/downloads/docs/devel-docs/

I’m no expert so I try and do my homework. Do your homework before arguing a point dude. It will really improve your chances of understanding.

I’ve learned a lot from this forum. The advice you are getting from ALucas and mikeiver1 sounds pretty reasonable in my opinion, but then again, you have your reasons for doing what you want. Whatever floats your boat dude.

Actually, XP is a free trial for 30 days - until you activate it. I am not sure about MacOSX. Haven't run it for ages.

Sanlite would be great. Windows 2003 server has the initiator. If you can boot up the server, forget about all the file sharing stuff, just have the iSCSI support, it can be very useful.

The reason to have 1 installer for everything is so people don't buy one version then figure out they need another. Please check how freenas does it. it doesn't complicate anything even for the dummies.

by the way, will there be a version to install it to the HD? I want to use it on small form PC for my home server which only has one HD and no floppy (and doesn't boot from USB). SFF PC can be ideal NAS because it is QUIET, take little space, and use little power. My E-PC is just a little bit bigger than my external CD writer.

(now to dimension)
The link you gave shows exactly two sections under "FreeNAS in details (TO BE FINISH)" - Initial startup script and Web Interface.

Like I said, I may not wrong. That gave me the impression that's all they wrote. Everything else is from different open source packages.

And by the way, please calm down. I don't even see Tony or Ralph get so excited when someone evaluate their products and point out problems (whether they agree or not).

mike, the level of RAID preference is mostly personal opinions. I don't like RAID-5 or RAID-6's rebuilding time. I have never used RAID-5 for critical data. RAID-1 ends up with a lot more drives and I am completely aware of that, and that's a cost I am willing to take.

OK, guess you know something about RAID5/6 that major financial institutions don't. I will say that if your experiance with RAID5 is based on a software implementaion then you may want to look at it again. Also note that there is still a rebuild time for a raid1 implemention just as there is with RAID5/6.

More to ponder,

Mike

SANLite gets my vote! Bring on iSCSI and AoE.

There is no ruling that all financial institutions would use one RAID or another. It is all up to the IT department.

Again, this is getting off-topic, since RAID preference is all individual decisions, but for some interesting reading, this is a link which had a great influence on me - http://www.baarf.com/

I will have to say RAID-6 is better and is not completely out of question yet. However, using dedicated RAID box is a completely different path and I would rather not discuss it in the NAS discussion, so let's leave it as "it is a personal choice".

Oops, I need to apologize. I misread the iSCSI feature in FreeNAS (no wonder I caused some confusion).

I thought it supports iSCSI target. Instead, it is an iSCSI initiator. That's useless for me.

What I wished was iSCSI target support - to make the drive used by some other servers.

Well, hopefully Naslite is listening... (hint)

There are many different reasons to select one RAID implementation over another - you have presented the cost side of the equation, RAID 1 is considered more expensive because only 50% of the total storage capacity is available. With RAID 5, the available capacity starts at 66% and improves as more drives are added (for RAID 5 available capacity = drive capacity x <total number of drives -1>)

The flip side is performance and scalability - RAID 5 and 6 require considerably more overhead - a WRITE requires the stripe to be read, new parity calculated, and then the stripe rewritten - the number of disk I/O operations increases with the number of disks in the array - this becomes an important issue for transactional database operations which require large numbers of small read/write operations.

I know that there is large overhead with a large RAID array. He did say that performance was not an issue though. By the way John, great little article on RAID5 but what has it to do with your array which you state performance is not an issue for. Also keep in mind that the article is almost 7 years old and that at that time SCSI U2W (80MB/sec) was the fastest interface. Memory used in hardware RAID cards was god awfull expensive and 256MB was unheard of for the most part. The processors implemented in the Current RAID cards are sometimes faster than those in the machine they are being used in. I am currently building an array and will be using a RAID6 stripe with a hot spare. the card will more than keep up with the load of IOs. By the way, the upper end of 3Ware SATA RAID cards will sustain close to 400MB/sec in writes. Me thinks I would not be concerned with the performance side of a RAID5 inplementation on writes. The place I work in does have a whole bunch of arrays and almost every one is RAID5, the DB is RAID50 and it's a big bastard and is hit hard all day.

Off topic, I don't know about but I will let it rest at this point. It would seem that you and I may need to find a bit more current data from which to pull tidbits about RAID implementations. I simply made thouse points to address all of your needs, money, heat, energy, and rack space.

Good luck and hope you find what your looking for.

Mike