Andrew Pollack's Blog
Technology, Family, Entertainment, Politics, and Random Noise| Professional Services | Second Signal | Presentations | Andrew's Blog | Support |
DAOS (Attachements stored outside the NSF in a pool) after 1 week - The Vedict
By Andrew Pollack on 07/01/2008 at 08:46 AM EDTIt's a winner! After one week, the most important thing I can say about using DAOS on my mail file, is that I'd totally forgotten about it.
DAOS enabled servers use a file system data store as a shared pool for file attachments across many databases to prevent duplication, drastically reduce file storage space requirements, reduce disk fragmentation, and improve NSF file performance and reliability.
After one week of using DAOS, my mail file remains about half the size it was before, but otherwise I'm unable to tell the difference. My local replicas and replicas on other servers (including clusters) are not impacted. If anything, performance has improved.
re: DAOS (Attachements stored outside the NSF in a pool) after 1 week - The VedictBy Rob Wills on 07/01/2008 at 01:09 PM EDT
re: DAOS (Attachements stored outside the NSF in a pool) after 1 week - The VedictBy Kevin Pettitt on 07/01/2008 at 02:24 PM EDT
The first question asked when I mentioned this on a client site last week was
"isn't this just a repackaged version of the old 'shared mail'?" I don't even
remember the reasoning anymore but it long ago became a matter of faith to
never implement that feature. So the concern is understandable. If someone
could talk about that it would be very helpful.
Following on from that point, a successful case study of DAOS on a very large
site would be very comforting. If I recall correctly, DAOS is implemented on an
individual server basis, so you could have a clustered pair of servers with
DAOS running on only one, right? The idea being that if DAOS goes flaky on that
box, the other remains viable, but of course there may be ways for the
flakiness to transfer over. Thoughts welcome.
"isn't this just a repackaged version of the old 'shared mail'?" I don't even
remember the reasoning anymore but it long ago became a matter of faith to
never implement that feature. So the concern is understandable. If someone
could talk about that it would be very helpful.
Following on from that point, a successful case study of DAOS on a very large
site would be very comforting. If I recall correctly, DAOS is implemented on an
individual server basis, so you could have a clustered pair of servers with
DAOS running on only one, right? The idea being that if DAOS goes flaky on that
box, the other remains viable, but of course there may be ways for the
flakiness to transfer over. Thoughts welcome.
re: DAOS (Attachements stored outside the NSF in a pool) after 1 week - The VedictBy Andy Broyles on 07/01/2008 at 02:30 PM EDT
I have not had an opportunity to explore 8.5beta...can DAOS be used outside of
mail for other Notes applications?
mail for other Notes applications?
re: DAOS (Attachements stored outside the NSF in a pool) after 1 week - The VedictBy Martin on 07/02/2008 at 04:22 AM EDT
As I understand the article at
http://www.dominoblog.com/dominoblog/dblog.nsf/dx/reduce-storage-costs-with-domi
no-8.5 it is possible to use it just for mailboxes as the attachments are
stripped out at mail.box. But then it shouldn't work on cluster server or
replicas as there is no connection with mail.box.
http://www.dominoblog.com/dominoblog/dblog.nsf/dx/reduce-storage-costs-with-domi
no-8.5 it is possible to use it just for mailboxes as the attachments are
stripped out at mail.box. But then it shouldn't work on cluster server or
replicas as there is no connection with mail.box.
re: DAOS (Attachements stored outside the NSF in a pool) after 1 week - The VedictBy Erik Brooks on 07/01/2008 at 07:38 PM EDT
The way it is advertised, DAOS works across any Domino NSF on the server.
This should be a *HUGE* win across the board, though there are a few
disadvantages:
- If you want to backup individual DBs you absolutely must back up your DBs
with a Domino-aware, api-level backup tool. If you want to simply backup
everything then that's easier: backup the NSFs, backup the filestore directory
and you should be good.
- It's only as robust as your filesystem. If your filesystem balks at tons of
files you should explore DAOS carefully.
- I've heard that you can technically "leave" Domino's security. I.E. a
plaintext file attached to a Domino doc can end up as plaintext in the
filestore directory, susceptible to anybody with read access to the directory.
Personally I'm excited about it from an app dev standpoint. I can offload gigs
of attachments out of NSFs (on very expensive and fast arrays) onto a more cost
effective SAN.
This should be a *HUGE* win across the board, though there are a few
disadvantages:
- If you want to backup individual DBs you absolutely must back up your DBs
with a Domino-aware, api-level backup tool. If you want to simply backup
everything then that's easier: backup the NSFs, backup the filestore directory
and you should be good.
- It's only as robust as your filesystem. If your filesystem balks at tons of
files you should explore DAOS carefully.
- I've heard that you can technically "leave" Domino's security. I.E. a
plaintext file attached to a Domino doc can end up as plaintext in the
filestore directory, susceptible to anybody with read access to the directory.
Personally I'm excited about it from an app dev standpoint. I can offload gigs
of attachments out of NSFs (on very expensive and fast arrays) onto a more cost
effective SAN.
re: DAOS (Attachements stored outside the NSF in a pool) after 1 week - The VedictBy Sebastian Leen on 07/05/2008 at 08:16 PM EDT
What about backup? I use a standalone backup utility (not related to domino). I
guess it will not work since there is no way to see which atts belongs to which
DB thus there is no easy way to restore a DB with the related Atts.
Also, I was told it was unsecure, that if I go to the DAOS folder and I make a
double click on the cached file, I can see it!!!
I was also told that spams will cause problem with DAOS since internally a MIME
email can be stored as an internal att so DAOS will cache it... so I will have
millons of atts in the DAOS folder, not too good.
Also it's not easy to measure the savings since it removes the atts from the
NSF but puts them in the file system...
guess it will not work since there is no way to see which atts belongs to which
DB thus there is no easy way to restore a DB with the related Atts.
Also, I was told it was unsecure, that if I go to the DAOS folder and I make a
double click on the cached file, I can see it!!!
I was also told that spams will cause problem with DAOS since internally a MIME
email can be stored as an internal att so DAOS will cache it... so I will have
millons of atts in the DAOS folder, not too good.
Also it's not easy to measure the savings since it removes the atts from the
NSF but puts them in the file system...
wow, LOTS of misinformaton in that replyBy Andrew Pollack on 07/05/2008 at 08:37 PM EDT
backup:
There is a backup api in Domino that will allow Domino aware backup
software to do the backups including the necessary files. This also will allow
better incremental backups for nsf files. Alternately, you can back up the
DAOS store with your backup files.
security:
I don't know who told you this, but these decisions aren't even finalized
yet. Files aren't going to be just dumped into the file system unsecured. I
can't talk about details because of NDA -- but there's plenty of history on
IBM's commitment to security. Its a subject being fully covered in the design
partner discussions and I'm sure IBM either has or will be making statements
fairly soon on this.
spams:
There are many ways to combat this. the first is of course better spam
protection. The second is that you can set the minimum size for DAOS to pick
out files. Spams don't send 100k attachments, they send (at most) small image
files. Also, if you're planning to use DAOS, you're going to need to take
responsibility for picking a file system that can handle it.
Out of curiosity, where are you hearing all this?
Also, the email address you've left here isn't valid. That makes me wonder,
frankly, where you come from.
There is a backup api in Domino that will allow Domino aware backup
software to do the backups including the necessary files. This also will allow
better incremental backups for nsf files. Alternately, you can back up the
DAOS store with your backup files.
security:
I don't know who told you this, but these decisions aren't even finalized
yet. Files aren't going to be just dumped into the file system unsecured. I
can't talk about details because of NDA -- but there's plenty of history on
IBM's commitment to security. Its a subject being fully covered in the design
partner discussions and I'm sure IBM either has or will be making statements
fairly soon on this.
spams:
There are many ways to combat this. the first is of course better spam
protection. The second is that you can set the minimum size for DAOS to pick
out files. Spams don't send 100k attachments, they send (at most) small image
files. Also, if you're planning to use DAOS, you're going to need to take
responsibility for picking a file system that can handle it.
Out of curiosity, where are you hearing all this?
Also, the email address you've left here isn't valid. That makes me wonder,
frankly, where you come from.
re: wow, LOTS of misinformaton in that replyBy John possi on 07/07/2008 at 10:07 AM EDT
Thanks Andrew, I never put my email address since I don't want to receive spams.
Here are my findings:
1 Backup: So I will need to change my usual backup utility by another one
coded using the Domino API. OK, I got the answer.
2 Security: I was told by a colleage that did the tests... anyway I got the
answer.
3 SPAMS: I think you didn't catch the question. Suppose that I enable DAOS for
mail.box (as IBM recommends). Now I receive one spam email so it is created in
the mail.box. At this time the attachment is already put in the DAOS
repository... it's too late for my antispam! My antispam will delete the email,
OF COURSE, but the attachment will be left in the DAOS repository until the
next prune is performed (one month). The attachment is put in the DAOS
repository as soon as Domino saves the document in the mail.box... antispam
acts AFTER the document is put in the mail.box.
4 SPAM2 Reply of Reply: I agree that antispam usually don't contain
attachments but just few images. But the HTML portion of the email (that is,
the body itself) is usually large than 100 KB. This is not only for spams but
also for reply of reply... So you will think "Ok, but DAOS is just for
attachments, not for the body"... but this is WRONG: When you receive a MIME
email (an email from internet) holding a body having more than a give size (I
don't know but I think 100 KB is enough) then the HTML and "Plain text" portion
of the email are stored in internal attachment. You can make the test, send an
email containing lot of text and then when you receive it you will see that
there are two $file internal attachment: One for the HTML and another for the
"Plain text" representation. So it basically means that if my users receives
plenty of personal emails containing replies of replies, or spams then my DAOS
folder will become huge until next prune.
5 What about clusters? Suppose in one physical server I have my main server
and the cluster... will I have two DAOS folders? I guess so since DAOS is in a
per-server basis.
6 Will it really save space? Suppose that I have thousands of users that
receives SPAM and plenty of attachments... I enable DAOS for all DBs. Now the
size of the NSFs will of course be reduced but each MB I took from the NSF will
be left in my file system. So DAOS magic is that duplicate atts will not be
repeated so the storage is based on removing duplicate attachments... but on
the other side there will be millons of attachments that will be in the DAOS
folder that were deleted by the user in the NSF (because they were SPAM or
replies of replies) so until next prune is performed these attachments will be
there. Consequence: At a given time (e.g. one day before my prune day) the
storage using DAOS will be bigger than without using DAOS since the size of the
not-yet-removed attachments will be bigger than the size of the
duplicate-atts-savings. Of course prunning each day is not a choice.
Thanks.
Here are my findings:
1 Backup: So I will need to change my usual backup utility by another one
coded using the Domino API. OK, I got the answer.
2 Security: I was told by a colleage that did the tests... anyway I got the
answer.
3 SPAMS: I think you didn't catch the question. Suppose that I enable DAOS for
mail.box (as IBM recommends). Now I receive one spam email so it is created in
the mail.box. At this time the attachment is already put in the DAOS
repository... it's too late for my antispam! My antispam will delete the email,
OF COURSE, but the attachment will be left in the DAOS repository until the
next prune is performed (one month). The attachment is put in the DAOS
repository as soon as Domino saves the document in the mail.box... antispam
acts AFTER the document is put in the mail.box.
4 SPAM2 Reply of Reply: I agree that antispam usually don't contain
attachments but just few images. But the HTML portion of the email (that is,
the body itself) is usually large than 100 KB. This is not only for spams but
also for reply of reply... So you will think "Ok, but DAOS is just for
attachments, not for the body"... but this is WRONG: When you receive a MIME
email (an email from internet) holding a body having more than a give size (I
don't know but I think 100 KB is enough) then the HTML and "Plain text" portion
of the email are stored in internal attachment. You can make the test, send an
email containing lot of text and then when you receive it you will see that
there are two $file internal attachment: One for the HTML and another for the
"Plain text" representation. So it basically means that if my users receives
plenty of personal emails containing replies of replies, or spams then my DAOS
folder will become huge until next prune.
5 What about clusters? Suppose in one physical server I have my main server
and the cluster... will I have two DAOS folders? I guess so since DAOS is in a
per-server basis.
6 Will it really save space? Suppose that I have thousands of users that
receives SPAM and plenty of attachments... I enable DAOS for all DBs. Now the
size of the NSFs will of course be reduced but each MB I took from the NSF will
be left in my file system. So DAOS magic is that duplicate atts will not be
repeated so the storage is based on removing duplicate attachments... but on
the other side there will be millons of attachments that will be in the DAOS
folder that were deleted by the user in the NSF (because they were SPAM or
replies of replies) so until next prune is performed these attachments will be
there. Consequence: At a given time (e.g. one day before my prune day) the
storage using DAOS will be bigger than without using DAOS since the size of the
not-yet-removed attachments will be bigger than the size of the
duplicate-atts-savings. Of course prunning each day is not a choice.
Thanks.
Let me try to answer some of these....By Andrew Pollack on 07/07/2008 at 11:03 AM EDT
...Though I have to say, I'm tempted to not bother given your propensity for
false input on the comment stuff. You know who I am, you should have courtesy
to tell me who you are when we're talking. Email address is not published
here, but is required. By my own rules, I should be pulling your comments.
The only reason I haven't, is because they're valid and interesting so far.
1. Backup: I don't know what your usual backup utility is so I can't answer.
DAOS is an enterprise solution, not a home server solution. If you're
seriously backing up massive domino mail directories with a $59.00 tool, I'm
not sure why I'm bothering to talk to you. The other major market tools are
being approached to participate in the API and I suspect support will be fairly
rich. Remember, we're still in early beta with this thing.
2. Security: Your colleage is right for the version currently being tested. I
think if you read the release notes for the beta, this is discussed. I know
for a fact that it won't be released this way. I can't say what the actual
plan is wrt encryption on disk and so on, but perhaps one of the ibm people who
read here will comment.
3. Spams: Why is your primary mail server your front end anti-spam device?
Sounds pretty small time to me. Maybe your environment isn't well suited to
using DAOS. Its a feature, not a requirement. My anti-spam happends long
before any mail reaches my user mail servers, and I'm a tiny shop. I know of
no major sites beyond the SOHO where spam gets all the way to mail servers
before anti-spam is done.
4. I think you'll find this isn't strictly true, but I'm going to have to let
someone from IBM answer it for now if they're willing.
5. At least at this time, DAOS is per-server. Domino clusters are totally
independent of each other. You can take an ax to one and the other will be
fine. You can cluster across wide area networks if you want. This is a very
different idea than the pseudo clustering Microsoft does. I've heard
discussions around the idea of a cluster using a shared DAOS store at some
point in the future -- perhaps if its being stored on a high availability SAN
where redundancy is handled outside of Domino. At present this isn't what
they're doing. I don't know if it will ever happen. I guess demand will drive
it.
6. Will it really save space? Your mileage may vary. Look into IBM's test
data (which I won't quote because I don't track what is NDA and what isn't) and
you'll see that it saves a huge amount of space for a majority of sites. How
much will be dependent on the way mail is used in the environment and other
factors but I know of no cases where its been tried and there has not been a
sizable space savings.
From everything I can tell so far, DAOS is far far from what the old crappy
"Shared mail" thing was. The version in Beta (and probably the initial stuff
in 8.5) barely touches the future potential of what it can do.
false input on the comment stuff. You know who I am, you should have courtesy
to tell me who you are when we're talking. Email address is not published
here, but is required. By my own rules, I should be pulling your comments.
The only reason I haven't, is because they're valid and interesting so far.
1. Backup: I don't know what your usual backup utility is so I can't answer.
DAOS is an enterprise solution, not a home server solution. If you're
seriously backing up massive domino mail directories with a $59.00 tool, I'm
not sure why I'm bothering to talk to you. The other major market tools are
being approached to participate in the API and I suspect support will be fairly
rich. Remember, we're still in early beta with this thing.
2. Security: Your colleage is right for the version currently being tested. I
think if you read the release notes for the beta, this is discussed. I know
for a fact that it won't be released this way. I can't say what the actual
plan is wrt encryption on disk and so on, but perhaps one of the ibm people who
read here will comment.
3. Spams: Why is your primary mail server your front end anti-spam device?
Sounds pretty small time to me. Maybe your environment isn't well suited to
using DAOS. Its a feature, not a requirement. My anti-spam happends long
before any mail reaches my user mail servers, and I'm a tiny shop. I know of
no major sites beyond the SOHO where spam gets all the way to mail servers
before anti-spam is done.
4. I think you'll find this isn't strictly true, but I'm going to have to let
someone from IBM answer it for now if they're willing.
5. At least at this time, DAOS is per-server. Domino clusters are totally
independent of each other. You can take an ax to one and the other will be
fine. You can cluster across wide area networks if you want. This is a very
different idea than the pseudo clustering Microsoft does. I've heard
discussions around the idea of a cluster using a shared DAOS store at some
point in the future -- perhaps if its being stored on a high availability SAN
where redundancy is handled outside of Domino. At present this isn't what
they're doing. I don't know if it will ever happen. I guess demand will drive
it.
6. Will it really save space? Your mileage may vary. Look into IBM's test
data (which I won't quote because I don't track what is NDA and what isn't) and
you'll see that it saves a huge amount of space for a majority of sites. How
much will be dependent on the way mail is used in the environment and other
factors but I know of no cases where its been tried and there has not been a
sizable space savings.
From everything I can tell so far, DAOS is far far from what the old crappy
"Shared mail" thing was. The version in Beta (and probably the initial stuff
in 8.5) barely touches the future potential of what it can do.
Other Recent Stories...
- 01/26/2023Better Running VirtualBox or VMWARE Virtual Machines on Windows 10+ Forgive me, Reader, for I have sinned. I has been nearly 3 years since my last blog entry. The truth is, I haven't had much to say that was worthy of more than a basic social media post -- until today. For my current work, I was assigned a new laptop. It's a real powerhouse machine with 14 processor cores and 64 gigs of ram. It should be perfect for running my development environment in a virtual machine, but it wasn't. VirtualBox was barely starting, and no matter how many features I turned off, it could ......
- 04/04/2020How many Ventilators for the price of those tanks the Pentagon didn't even want?This goes WAY beyond Trump or Obama. This is decades of poor planning and poor use of funds. Certainly it should have been addressed in the Trump, Obama, Bush, Clinton, Bush, and Reagan administrations -- all of which were well aware of the implications of a pandemic. I want a military prepared to help us, not just hurt other people. As an American I expect that with the ridiculous funding of our military might, we are prepared for damn near everything. Not just killing people and breaking things, but ......
- 01/28/2020Copyright Troll WarningThere's a copyright troll firm that has automated reverse-image searches and goes around looking for any posted images that they can make a quick copyright claim on. This is not quite a scam because it's technically legal, but it's run very much like a scam. This company works with a few "clients" that have vast repositories of copyrighted images. The trolls do a reverse web search on those images looking for hits. When they find one on a site that looks like someone they can scare, they work it like ......
- 03/26/2019Undestanding how OAUTH scopes will bring the concept of APPS to your Domino server
- 02/05/2019Toro Yard Equipment - Not really a premium brand as far as I am concerned
- 10/08/2018Will you be at the NYC Launch Event for HCL Domino v10 -- Find me!
- 09/04/2018With two big projects on hold, I suddenly find myself very available for new short and long term projects.
- 07/13/2018Who is HCL and why is it a good thing that they are now the ones behind Notes and Domino?
- 03/21/2018Domino Apps on IOS is a Game Changer. Quit holding back.
- 02/15/2018Andrew’s Proposed Gun Laws
servers with lots of mail files because of the duplicated attachments sent to
numerous recipients. This should significantly reduce the e-mail storage costs
of large corporations using Domino. Rob