DHCP/Router Quad Port Nics on Same Server Challenging Question
How do I make this work?
On the two built in ethernets, one is used for the net which has the following settings:
Gateway: 192.168.2.1
IP Address: 192.168.2.10 (Servers Primary IP)
Subnet 255.255.255.0
My DHCP Scope is
192.168.2.25 - 192.168.2.125
255.255.255.0
Default Gateway 192.168.2.1
Never works, any help?
Reply:
What it does give you is four NICs in the host. Each port needs to be plugged into a switch port on one or more external switches to give the server four network connections.
To use this machine as a DHCP server and router you will need to plug one port into a network which connects to the Internet. You will need to plug another port into the same switch as your client machines connect to.
Bill
------------------------------------
Windows Vista to XP downgrade complete - and here's why
Perhaps my first and foremost complaint with Vista was the trend towards friendliness to the point of frustration.
The first example of this is the integration of Windows Defender. I find even XP's defender irksome. I understand that there are a great many people out there who do not know how to keep their Windows secure. I am not one of them, and to assume I am on a highly frequent basis is quite frankly insulting.
The second of over-friendliness is your new search system. The length of time to realize how quirky it had become was pretty embarssing. One night I was searching for a file, but it was nothing important, and I was distracted. When no results were found, I assumed I hadn't copied the file from my server and that was that. An hour later, I had a nagging feeling I was certain I had copied the file. I browsed my folders manually and found it. Interesting, but still distracted, so the quirk was put out of mind.
Some weeks later, I was to transfer some files to a client. I copied them from my server to my laptop in the morning, and in the afternoon, when it was time to transfer, I searched for the files and found - no results. Interesting? I again found the files manually, but this time returned to the problem at once. At which point I realized one must use the "Advanced Search" in order to find a file not already opened on the system. Not interesting. Just annoying.
It was but a day or two later when I discovered yet another of the angrily "friendly" features. The way wireless networks are stored is absolutely boggling. It would seem I must elucidate you to the fact that people who own laptops with wireless network cards may tend to use those very same laptops in a number of locations on a number of networks. That one must disable and re-enable their wireless device in order to refresh the visible networks in a given area is absolutely idiotic.
Finally; Vista is simply slower. Vista Business runs slower than XP Professional on the same computer in a number of areas. I am highly familiar with services and startup options in the registry, and so it is not due to those factors. Running as minimal of services as possible on both OS's: it takes longer for Vista to boot from logon screen to usability, it takes longer for windows explorer to refresh views, and a number of programs (whether set to Vista-mode or optimized for XP) run at least a small percentage slower.
To tell the truth, that wasn't the final complaint. After being insulted, annoyed, played dumb and slow with, I then discovered Vista cannot go away. Technically the downgrade is not complete; with no floppy on my laptop, no XP bootdisks can be made. So five gigabytes are henceforth a permanent blackhole on the drive. Storage is cheap, but it's a constant reminder of the "user-friendly" attitudes I'd tried so hard to avoid.
Yet one final insult remains. I convinced a friend to invest in your company a few months prior to the release of Vista. If you travel back in time to any point in the last two to twenty years, and tell my younger self I would convince a friend to invest in Microsoft in the future, I would have slapped myself in the face then and there. While Windows has always been the OS of choice, your name has been decried on many more than a few occasions. I actually believed in a new Microsoft OS for the first time since Win95. That is part of why I waited so long, and gave Vista so many chances, before finally resolving to reinstall XP. I have the Ubuntu install CD nearby, though I've yet to install it. Perhaps, soon.
Good day.
(To view my application for CEO, view my profile.)
Reply:
Hi Danny,
Thank you for the post.
We always welcome customer feedback and thank you for providing us with your comments. If you have any concerns regarding Microsoft products, please contact our customer service. We strive to capture any and all product issues/feedback so as to ensure we are continuously developing Microsoft products to meet customer needs. You are also welcome to add your comments to make Microsoft products easier and more powerful to use.
To submit additional product feedback, we recommend that you visit http://www.microsoft.com/mscorp/execmail/feedback.mspx .
Tim Quan - MSFT
------------------------------------
Add a Minimize button to remote desktop when maximized
Graham
graham
Reply:
Matt Fraser, STO Forums Software Developer
------------------------------------
Bug: Memory leak on large network file copies, Server 2008-RTM
When I copy large VHD's (35 GB) between two servers in two different workgroups using Windows Explorer, memory usage climbs until all the memory is used up. When the copy completes, the memory is not released. I just installed Hyper-V RC0 and am now noticing the problem. Server has 24 GB RAM. When I start server, only 3GB RAM is used, during large network file copy, all available RAM disappears and is never released. Need to reboot server to free up the RAM. This is repeatable.
Reply:
Hello Peter -
We too are having this issue, can you read over
http://forums.microsoft.com/TechNet/ShowPost.aspx?PostID=3043575&SiteID=17
Are you experiencing the same issue? Any headway?
Brett
------------------------------------
Reply:
As a workaround, I now use ROBOCOPY from the command line to copy files. This results in all the FREE memory going to cache, so free memory goes to zero and cache memory increases, but overall memory usage does not go up and the server continues to function. I can't really experiment further as this server is a production box (hyper-v role) hosting four virtual servers and I can't risk bringing the server down.
Good luck,
-Peter
------------------------------------
Reply:
UPDATE:
I have set up a second Windows 2008 server, also with hyper-v role, but with different hardware and the copy problem does not happen. The memory still gets used up during the copy, but it gets released when the copy finishes.
The server which has the problem is using a 10GbE NIC. I'm thinking there may be some issue with the NIC driver or some of the TCP offload functions. I'll be investigating further and post my findings.
-Peter
------------------------------------
Reply:
Any progress?
Thanks
------------------------------------
Error Installing MIIS SP2 with SQL 2005
I'm trying to install MIIS SP2 onto a machine that has a local install of SQL 2005. I keep getting the following error no matter what I try:
"The service account cannot access SQL Server. Ensure that the server is accessible, the service account is not a local account being used with a remote SQL Server, and that the account doesn't already have a SQL login."
I'm signed onto the machine with a domain account that has a login account created in SQL and has been added to all of the SQL roles including sysadmin. I'm also using a domain account to run the MIIS service as, which I've tried doing the install without a SQL login and then again with one.
Any suggestions?
Reply:
Hello Ken
This is most likely as a result of permissions. The account that connects to the SQL server in order to create the DB is NOT the currently logged on account, but the SERVICE ACCOUNT that ILM runs under. Thus the ILM Service account needs permission to the database.
Just for the installation process the Serivce account needs "db create" permissions (which I usually just get by giving the account sysadmin). After the DB is created the role is removed and I give the account DBowner (if the DBA's) are ok with that.
HTH
-Almero
------------------------------------
Reply:
This is all transpiring on virtual machines and here is what I've tried so far:
Domain KZ on Virtual Server KZVS1
- miisadmin account added to domain admins
- SQL Login created and account added to all SQL roles
- Administrator account added to domain admins
- SQL Login created and account added to all SQL roles
SQL 2005 on Vistural Server KZVS2
- Local Administrator Account
- Added to the local administrator account
- SQL login created and account added to all SQL roles
Here is the combination of signon/miis service installation attempts. "Signon" indicates the windows account that I signed on to the server with and "miis service" is the account I used to setup the miis services (these were typed into the setup wizard for MIIS). All of these ended with the same error!
KZ\Administrator signed on - KZ\miisadmin for service setup
KZ\Administrator signed on - KZ\Administrator for service setup
KZ\miisadmin signed on - KZ\miisadmij for service setup
KZ\miisadmin signed on - KZ\Administrator for service setup
KZVS2\Administrator signed on KZVS2\Administrator for service setup
KZVS2\Administrator signed on KZ\Administrator for service setup
KZVS2\Administrator signed on KZ\miisadmin for service setup
etc.
------------------------------------
Reply:
Have you simply tried to install MIIS and specify the Service Account used to install SQL 2005? See if you get the same issues.
------------------------------------
Reply:
I'm getting the same problem, won't work with any of
SQL server 2005 Enterprise (remote)
SQL server 2000 (local)
SQL server 2005 Express (local)
Using the same account for SQL and MIIS. I've maybe tried 100 different combinations of instances, servers and usernames....
Using Windows 2003 Enterprise R2 SP2 32-bit.
------------------------------------
Reply:
Did anyone ever figure this out? I am getting this same error. I have installed many instances of ILM 2007 sp2 in the past with no issues. I am working with a new dba now and I am running into this problem. I had the dba give the ILM service account the sysadmin role and I still get the same error.
Any ideas would be much appreciated!
------------------------------------
Reply:
I am having this problem as well and was wondering if anyone had come up with a solution yet.
------------------------------------
Reply:
For me, the problem was resolved with I typed used "Administrator" (note the capital 'A') for the username and "DOMAIN" (note the domain in all caps) for the domain. Perhaps it is case sensitive?
------------------------------------
Configuring MS Search to Crwal Custom URL Types Using HTML IFilters
I need to crawl our Intranet Portal Landing Page (Implemented in Bea Portal) via MS earch Express. The page extension is ".pt" as opposed to .html, etc. so, when I try to add a new content source (under Search Server SSP) to point to our Intranet Portal Landing Page, I get an error, since the landing page in the portal is addressable by .../portal/server.pt?nm=xyz?cls=bsb....
To solve the problem, I tried to create a new extension (.pt) in my registry (steps shown below), and stoped/started search services via command line. But, I am still getting an error message. I would very much appreciater any help I can get here. Thanks.
Step 1: I defined the extension ".pt" in my reistry under the following key:
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Office Server\12.0\Search\Setup\Filters\.pt]
"Extension"="pt"
"FileTypeBucket"=dword:00000001
"MimeTypes"=<MIME Type>"
step 2: I associated html filter GUID with the extension . pt, so that when content source ends with .pt, search would use html filter.
[HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Office Server\12.0\Search\Setup\ContentIndexCommon\Filters\Extension\.pt]
(default)"="{56BD18AD-CF9C-4110-AAAA-B2F96887D123}"
Does anyone else has any other suggestion?? Thanks.
Backup all DBs of a SQL-Server Instance
The script allows to backup all databases if a SQL-Server Instance. It determine the backup-path by the environment variable DBBackupPath of the operating system. It also determines the SQL-Server instance on which it runs and uses the corresponding path.
1.) Create a Backup-Folder on the Server. The last part of the path must be "Backup"
(e.g. f:\MSSQL2005\Backup)
1a.) If there is a SQL-Server instance installed on the server the instancename must be previous before "Backup" (e.g. f:\MSSQL2005\Instance_Name\Backup
2.) Set the environment variable "DBBackupPath" to this folder
3.) Run the script by scheduler (ISQL/OSQL) or SQL-Query
declare @BackupPath nvarchar(512),
@DB nvarchar(512),
@Status Int,
@SQLCommand nvarchar(1024),
@szVersion char(1),
@EnvVarName nvarchar(256),
@ServerName nvarchar(64),
@InstanceName nvarchar(64)
-- For version 9 on free up xp_cmdshell
select @szVersion = Substring (@@version, CharIndex ('-',@@version) + 2,1)
if @szVersion > '8'
begin
exec sp_configure 'show advanced options', 1
reconfigure
exec sp_configure 'xp_cmdshell', 1
reconfigure
end
-- Get the environment variable DBBackupPath
-- The last part should be 'Backup'
set @EnvVarName = 'DBBackupPath'
set noCount on
create table #EnvirnmentVariables (output varchar(1000) NULL)
insert #EnvirnmentVariables exec master..xp_cmdshell 'set'
Select @BackupPath = Substring (output, CharIndex ('=',output) + 1, len(output) - NULLIF(CharIndex ('=',output),-1))
from #EnvirnmentVariables where
upper (@EnvVarName) = upper (Substring (output,1, NULLIF(CharIndex ('=',output)-1,-1)))
drop table #EnvirnmentVariables
select @InstanceName = ''
-- no instances in version 7 and below
if @szVersion > '7'
begin
create table #SQLServers (name varchar(128) NULL, network_name varchar(28) NULL, status varchar(100) NULL, id varchar(4) NULL, collation_name varchar(128) NULL, connect_timeout varchar(15) NULL, query_timeout varchar(13) NULL)
insert #SQLServers exec master..sp_helpserver
Select @InstanceName = upper (Substring (name,NULLIF(CharIndex ('\',name),-1),128))
from #SQLServers where id=0
drop table #SQLServers
if NULLIF(CharIndex ('\',@InstanceName),-1) = 0 select @InstanceName = ''
end
set noCount off
if @szVersion > '8'
begin
exec sp_configure 'xp_cmdshell', 0
reconfigure
end
select @BackupPath = Substring(@BackupPath,1,Len(@BackupPath)-7) + @InstanceName + N'\Backup'
print 'Backup Path:'
print @BackupPath
print 'SQL-Server Major Version:'
print @szVersion
print 'SQL-Server Instance:'
print @InstanceName
-- DB-backup...
DECLARE DBCursor CURSOR FAST_FORWARD FOR
SELECT name, status FROM master..sysdatabases
OPEN DBCursor
FETCH NEXT FROM DBCursor INTO @DB, @Status
WHILE @@FETCH_STATUS <> -1
BEGIN
IF @DB <> 'distribution' and
@DB <> 'tempdb' and
@DB <> 'Northwind' and
@DB <> 'AdventureWorks' and
@DB <> 'pubs'
begin
print 'Backing up database: ' + @DB
IF @Status & 8 = 0
begin
print 'Log...'
SELECT @SQLCommand = N'Backup log [' + @DB + '] to disk = N' + char(39) + @BackupPath + char(92) + @DB + 'DumpLog.bak' + char(39) + ' with init'
EXEC (@SQLCommand)
end
print 'Data...'
SELECT @SQLCommand = N'Backup database [' + @DB + '] to disk = N' + char(39) + @BackupPath + char(92) + @DB + 'DumpDat.bak' + char(39) + ' with init'
EXEC (@SQLCommand)
end
FETCH NEXT FROM DBCursor INTO @DB, @Status
END
DEALLOCATE DBCursor
Reply:
------------------------------------
Windows did not start successfully
I've just returned from a weeks holiday, switched PC on and hit problems - hope someone can help?
Running Windows xp with SP2, but not getting the operating system loaded!
I first got a 'No signal input to monitor' displayed and have seen this before so done same as preiously and disconnected/reconnected monitor connector from PC, then rebooted.
'Windows did not start properly' screen came up but I have yet to continue beyond here....the timer does it's countdown to 0 then does nothing, or if I make a selection and press 'return' or 'enter', the screen freezes with the timer stopping on whatever number of seconds it reached!!
Can you help?
Reply:
Hi Stevie,
Can we enter Last Known Good Mode or any other mode when the computer starts?
------------------------
1. Restart your computer and start pressing the F8 key on your keyboard.
2. When the Windows Advanced Options menu appears, select Last Known Good Mode, and then press Enter.
Note: If this mode doesn't work, try Safe Mode also.
Please understand that this is a general Windows XP startup problem. As this forum mainly focus on SteadyState related issues, this inquiry would best be posted to Windows XP public newsgroup:
The reason why we recommend posting appropriately is that you will get the most qualified pool of respondents, and other customers who read the newsgroups regularly can either share their knowledge or learn from your interaction with us. Thank you for your understanding.
Best Regards,
------------------------------------
Reply:
------------------------------------
Fix layout
------------------------------------------------------------------------------------------
Browser title
Browser menu bar
Browser function bar
Browser quick links bar
Browser second quick links bar
Browser tabs bar
Login & Microsoft links
MSDN title and MSDN Search box,
which is thick and takes two rows
MSDN Forums title and user icon,
which is thick and takes two rows
Function buttons and user name
White space and user medals
White space
Forum category links
Forum title, search box, and help link
Forum description and Browse by Tag -title
Announcements title and Browse by Tag box
Announcements thread link
Some white space around boxes
First row of threads functions box
Second row of threads functions box
Some white space
Page numbers and thread list field titles
1st visible thread and Statistics title
Post information about 1st visible thread first row plus white space and Statistics box
Post information about 1st visible thread second row plus white space
2nd visible thread
Post information about 2nd visible thread first row plus white space
Post information about 2nd visible thread second row plus white space and Top Answerers title
3rd visible thread and Top Answerers box
Post information about 2nd visible thread first row plus white space
Post information about 2nd visible thread second row plus white space
Browser status bar
------------------------------------------------------------------------------------------
I can try to reduce my browser bars, but that is not enough alone. It would be better if you removed white space, combined some function bars, removed / moved unnecessary things somewhere else, and combined "information about the thread"-rows into a single row.
The old forums is bad. There is lots of bugs and layout was almost unusable until I adjusted my screen resolution bigger. After that I see 7 threads visible at first after loading the page. Now in the new forums I see only 3 threads visible at first. So its more than half worse than the old forums. Though the old forums still has bad layout on "information about the thread", because it creates more unnecessary rows even when there is white space to use, if I make the browser window smaller horizontally.
When I browsed some more, I saw more threads at first after downloading the page, because the announcements box is shown only at the first page. But I still see only 4 and half threads at a time until I scroll down. It goes to top of the page automatically too instead of going to top of the threads list.
I should be able to see more thread titles without scrolling down. The thread titles are the main content, not everything else.
Also in the subcategory page there is lots of unnecessary white space.
- Edited by Silvercode Monday, August 18, 2008 7:42 AM Added links
Exchange Server Documentation Updates - August 2008
Here are the latest update to Exchange server content from Exchange Server documentation team:
http://msexchangeteam.com/archive/2008/08/13/449576.aspx
I highly recommend you to review the latest update contents on Exchange server 2007 with Service Pack 1. There are couple of new updates and several good white papers. Enjoy it! 
Tiandameinv - flood
http://forums.msdn.microsoft.com/en-US/user/Tiandameinv/threads
Someone please take a look at this
Lucian Baciu, MCTS, http://studentclub.ro/lucians_weblog
Reply:
Matt Fraser, STO Forums Software Developer
------------------------------------
Reply:
Lucian Baciu, MCTS, http://studentclub.ro/lucians_weblog
------------------------------------
Reply:
Hans Passant.
------------------------------------
Reply:
The Vote + Post is abusive feature is quite dysfunctional. Moderators can only find them back by applying a filter for each individual forum. There are not nearly enough abusive posts in any single forum to make that worth the effort and there are far too many forums with far too few moderators to make this a daily hassle. The "Reporting Forums" forum at the old site worked much better. Push vs pull. I'm not bothering with it in the forums on this site, nobody else is either as near as I can tell.
Hans Passant.
Ok, then something has to be done about this.
Lucian Baciu, MCTS, http://studentclub.ro/lucians_weblog
------------------------------------
Empowered by SharePoint
To view the rest of this blog entry, visit http://bit.ly/7ESwh
Reply:
------------------------------------
The new MSDN Forums
Reply:
Once you get to the reply pages, it is definitely an enhancement - especially the possibility to see the entier thread that you are replying to. However, the front page is awful. Low contrast and filled up with nonsense information like the Browse-by-tag box and the slowly blending medals and top answerers. It is also very irritating that you cannot use the forum with IE safety set to high - even if you add the forum to the trusted sites with lower safety, and we still miss a button, which can bring us back to the thread list after replying to a thread.
It is exactly like Vista. Not much gained, but flickering and slowed down to a level where it is no longer fun to visit.
It is actually funny to see the success of the very cheap Asus eeePC's. They are not able to run Vista, so Microsoft has a choice: Continue to support XP or leave the school market to Linux. Read my signature Microsoft. In the long run you need to learn to do things simple and efficient if you want to survive. The next step is a complete PC in a mobile phone, which can then be connected to a full-size screen, keyboard and mouse when you get home (see http://www.innovatic.dk/iteng.htm ).
A funny little question to think about:
How much electrical power (not CPU speed etc.) is necessary to make a PC run Vista fluent?
How much electrical power is available in a small mobile phone?
Sorry, a little off topic maybe, but the analogy to this forum is striking.
Everything should be made as simple as possible, but not simpler.
- Edited by Carsten Kanstrup Saturday, August 16, 2008 8:36 PM addition
------------------------------------
Reply:
Also I would like to have subcategories so that languages were in one (VB6, C#, ASP, ...), technologies in one (.Net, BizTalk, SQL Server, ...) architecture and design in one (general, role, ...), and so on or such.
Plus why there are such stupid categories that have title like Development and then they concentrate on some specific technology or platform only? I think if the title is general, the forum should be general too. So why under development is not eg. SQL Server development? Is SQL Server development less important? Or why there is not software development, which is general?
Performance is still a major issue, because when I clicked Submit, nothing happened. I clicked several times more and nothing happened. Then after 3 seconds the page suddenly submitted. And I really hate when I move to next page on the thread list and the page goes first to top, waits 2 seconds and refreshes. It should first refresh and then move back to top. Some other bugs are not fixed either.
- Edited by Silvercode Monday, August 18, 2008 6:42 AM Added performance
------------------------------------
The Name attribute of a Field is only 32 characters long???
For example for the Field definition:
<Field ID="{959B07DE-A6B8-4c05-8E52-82588F8164CB}"
Name="April_x0020_Actual_x0020_Quantitative"
SourceID="{51669f5c-7d34-489b-9f72-ca63c1dcc040}"
StaticName="April_x0020_Actual_x0020_Quantitative"
Group="DY Consulting"
Type="Number"
DisplayName="April Actual Quantitative" />
is accepted and installed fine!
But when inspected using the GetList web service we see that the Name attribute is returned as Name="April_x0020_Actual_x0020_Quantit"
Seems a bit odd (BUG?)
-Gavin
Reply:
<Field Type="Text" DisplayName="April Actual Quantitative" Required="FALSE" MaxLength="255" ID="{6cdcd3aa-88cc-40a0-8b56-e5b1aaf7339e}" SourceID="{cd9a3ba7-ed05-4cbc-b08e-75daaab01744}" StaticName="April_x0020_Actual_x0020_Quantitative" Name="April_x0020_Actual_x0020_Quantitative" ColName="nvarchar20" RowOrdinal="0" />
Is it possible that you should not have the _x0020_ in place of the spaces in the field definition when installing?
http://www.certdev.com
------------------------------------
Reply:
That's well under 32 Characters.....
Could you try using something OVER 32 Characters?
(plus I didn't set this up, it's been inherited from another dev)
------------------------------------
Reply:
http://www.certdev.com
------------------------------------
Authentication for client applications accessing SSDS resources
I'm starting this discussion again here at the request of Mike.
Basically, I wanted to know if there was, at the very least, a "yes or no" answer to the question..."Will client applications that make SSDS requests be able to do so securely (directly to my Containers) without having to go through a central server owned by the app developer?"
If the answer is "yes", that would be enough for me to happily rip out all my current source code and start afresh. The answer to the question is important in that it pretty much determines how we end up architecting our applications.
Funneling requests through a secured central server pretty much limits our ability as devs to scale our applications easily...which is kind of the point of SSDS (ok, ONE of them...but an important one)
So SSDS team...would it be possible to give us just a simple "YES" or "NO" on the issue?
Also, if anyone has been thinking about this and found a suitable workaround...lets use this thread to discuss architecture.
Thanks :)
- Edited by Senkwe Chanda Monday, August 11, 2008 5:59 PM Added hyperlink
Reply:
[Thanks for elevating the thread. I didn't want to see it buried at the end of my post on my demo code].
I'll let SSDS team speak to the question directly. I'll offer, however, my current angle on this for the sake of discussion. I currently am using a central Server Proxy to handle all SSDS requests. I do this for a couple reasons:
- I have some needs/wants for a web-based client-server conversation that SSDS is not currently doing (more extensive caching, compression, general redirection via 303, etc.)
- I want to expose the data I have in my SSDS stores to anonymous (non-SSDS auth'ed) users. I actually use SSDS as a user-profile store and authorize users against that.
In some cases, SSDS will (at some point) add one or more of the behaviors I am implementing w/ my proxy. That's ok w/ me. When SSDS has the new feature, I can update my proxy to remove that overlap and clients are not harmed. I've done this for a few things already. In other cases, the feature I implement is non-SSDS related (controlling unique ID generation, determining if POST is allowed, but not PUT, handling deleting a collection of records with a single DELETE request, etc.) and I'll continue to use a proxy to handle these tasks.
I will also say that, currently, I am thinking of SSDS as a passive (but smart) data store - not as an (inter)active data server. In other words, I don't expect to be able to *program* the SSDS servers w/ business rules, context validations, and other application-level work. Instead, I plan on doing with with server-side code that could be deployed on one or more servers.
I get some sense, reading this forum, that others are not of the same mind. That they view SSDS as a full-fledged application environment that would handle authentication/authorization, computation (SUM, AVG, GROUP-BY, etc.), and possibly even workflow. Not sure how far down that trail SSDS will be going in the near term. And would love to hear SSDS team feedback on that last item.
So there you have it. I'm embracing the proxy model (like I do w/ current SQL Server work).
Mike Amundsen [http://amundsen.com/blog/]
------------------------------------
Reply:
Senkwe Chanda:
[Thanks for elevating the thread. I didn't want to see it buried at the end of my post on my demo code].
I'll let SSDS team speak to the question directly. I'll offer, however, my current angle on this for the sake of discussion. I currently am using a central Server Proxy to handle all SSDS requests. I do this for a couple reasons:
- I have some needs/wants for a web-based client-server conversation that SSDS is not currently doing (more extensive caching, compression, general redirection via 303, etc.)
- I want to expose the data I have in my SSDS stores to anonymous (non-SSDS auth'ed) users. I actually use SSDS as a user-profile store and authorize users against that.
In some cases, SSDS will (at some point) add one or more of the behaviors I am implementing w/ my proxy. That's ok w/ me. When SSDS has the new feature, I can update my proxy to remove that overlap and clients are not harmed. I've done this for a few things already. In other cases, the feature I implement is non-SSDS related (controlling unique ID generation, determining if POST is allowed, but not PUT, handling deleting a collection of records with a single DELETE request, etc.) and I'll continue to use a proxy to handle these tasks.
I will also say that, currently, I am thinking of SSDS as a passive (but smart) data store - not as an (inter)active data server. In other words, I don't expect to be able to *program* the SSDS servers w/ business rules, context validations, and other application-level work. Instead, I plan on doing with with server-side code that could be deployed on one or more servers.
I get some sense, reading this forum, that others are not of the same mind. That they view SSDS as a full-fledged application environment that would handle authentication/authorization, computation (SUM, AVG, GROUP-BY, etc.), and possibly even workflow. Not sure how far down that trail SSDS will be going in the near term. And would love to hear SSDS team feedback on that last item.
So there you have it. I'm embracing the proxy model (like I do w/ current SQL Server work).
Mike Amundsen [http://amundsen.com/blog/]
Mike, I'm pretty much on the same wavelength as you regarding what SSDS is. I also don't expect it have a "programmable interface". I'm fine with it not supporting DB specific functionality. For me the major selling point of SSDS is that it alleviates me from having to administer my own database server(s) and worry about backup and recovery and scaling out and performance tuning etc. I'm not good at that. And even if I was, I wouldn't want it to get in the way of writing my application. But in your scenario, presumably you now have to worry about what happens to your server should you start approaching millions of requests via your proxy. I assume you also incur the cost of requests to/from your server to/from SSDS and back to the client...yes? I think the cost savings of having authenticated clients grab resources directly from my containers could be considerable beyond a certain point.
*fingers crossed*
------------------------------------
Reply:
*fingers crossed*
It is my *assumption* that bandwidth and request count will be billable items once SSDS goes live. That's one of the reasons I'm spending some time building a proxy that can do some local caching and filtering of SSDS requests. I currently can reduce the number of requests I make to SSDS this way and expect that this will continue to be the case in the future. This is especially handy for 'anonymous' requests from lots of users. In some cases I can do the authorization locally and pass them the cached shared copy of the resource w/o out involving SSDS. Again, I think we're on the same wave length here.
As for the proxy handling millions of requests. This challenge is no different than a web app built on SQL Server data. I happen to be focusing on a very small stand-alone proxy codebase that could be deployed in a web farm itself. That way I plan to be able to scale-up the ability to handle SSDS requests. Since all my work to date as been on just tens and hundreds of users/records, I can't really speak to the viability of this pattern yet.
In recent posts here, I've gotten the sense that the SSDS team is considering a native authentication/authorization model for SSDS. However, even after some poking and prodding, I was not able to tease out just what that might look like. We'll have to see. Also, SSDS has been pretty responsive to requests for adding caching support details. I'm hopeful that this will continue to include caching for lists/queries, and even into supporting compression. All that would help w/ bandwidth and request traffic.
Mike Amundsen [http://amundsen.com/blog/]
------------------------------------
Reply:
So, the answer to your question is a bit more difficult then a simple yes or no. I say this because the decision about including a middle tier between SSDS depends quite heavily on the scenario that you're in. For many of the reasons that Mike layed out (caching, compression, etc) these are protocol level things and are things that are already on our plate of things to do. So, if these are the primary reasons for including a middle tier then I would say that the inclusion of the middle tier in your app architeture is likely to be temporary.
Now, for some of the other items mentioned (SUM, Group-By, Order by, etc) these are core DB things which we will likely implement within the service. The timing of these things will obviously vary depending on the difficultly, the barrier it presents to scaling (if any), and resource constraints. However, I would also view these as core things that SSDS should offer to its users and therefore these should not necessarily be a reason for introducing a middle tier to your app architecture unless it is a temporary thing.
There are however stronger reasons to introduce a middle tier to your app architecture. I would consider the following to all be valid reasons for doing this:
1) Advanced caching scenarios to meet scaling demands (beyond protocol level caching). This could be caching the, "hot" users of your system or caching what you know to be your hot data, etc.
2) Complicated business logic scenarios. NOTE: I would say that Mike's mention of complicated delete scenearios may fall in this category if it can't be accomplished by deleting the container. Another example would be just enforcing your own constraints. Perhaps you have a photo sight you want to limit users to a particular # of uploads. You get the idea.
3) Stronger authz and authn requirements (in others words there is a single, "service" user but n actual users of the app). The other authn/z comes from you middle tier ensuring that they have the necessary credentials to perform the operations that they are attempting.
NOTE: This list is by no means meant to exhaustive but these at least represent the reasons that I would think would lead one down that path.
Hope it helps,
--Jeff--
------------------------------------
Reply:
yeah, your response makes lots of sense (to me, anyhow<g>).
1 - proxying to 'cover' missing baseline features in SSDS
2 - proxying to add new (non-SSDS) features
3 - proxying to support custom workflow, security, etc.
fwiw, as i move forward w/ my SSDS-Proxy work, i'm doing my best to keep items #1 & #2 clearly separated. and i'm keeping item #3 out of the SSDS-Proxy and keeping that in a separate code-set.
finally, to make this process of keeping proxies up-to-date, it helps to know - as soon as possible - what features are on the horizon for SSDS. please keep that kind of information coming, eh?
Thanks for listenin'
Mike Amundsen [http://amundsen.com/blog/]
------------------------------------
Reply:
I hear you loud and clear on the what feature is coming in what sprint issue. I think Soumitra's recently blogged about something similar to this. I think one of the difficulities that we have at the moment is that we haven't really settled on the final feature set for a given sprint until just before we begin to enter the sprint in question. So, providing a list of what features are coming in each sprint can be somewhat problematic at the moment (at least beyond the current sprint in execution).
Now, having said that I think we're going to try to get a little bit more transparent (per Soumitra's blog) on these things but I think we're still asking for a bit of patience on this at the moment.
Regards,
--Jeff--
------------------------------------
Reply:
Mike Amundsen [http://amundsen.com/blog/]
------------------------------------
Unable to share disk resource in a Hyper-V cluster
Hi,
I'm currently testing System Center Virtual Machine Manager 2008 together with a 3 node Win2008 cluster running Hyper-V. A big pain is that each VM in the cluster need their own disk resource. This is a big limitation as every time a new VM shall be created we need to do as follow:
1) Contact the SAN administrator to assign new small LUN to the cluster nodes.
2) Create a disk cluster resource
3) Then create a VM from the SCVMM2008 console
First of all this is too much work and delay when provisioning new VM's (it should be possible to provision a VM from a single point) and secondly we are unable to utilize the SAN disk's as we will end up with a lot of unused disk in each cluster disk resource.
I understand this to be a limitation in NTFS or in the Microsoft Cluster technology (?) as only one cluster node can access the same disk resource. Are there any plan in the future to remove this limitation so that several VM's in the cluster can share the same cluster disk resource and that the cluster nodes can access the same disk simultaniously.
This is one of the biggest disadvantages with Hyper-V compared to a VMware HA environment where this work fine.
best regards
Lidvar Kornberg
Senior System Engineer
Det Norske Veritas
Reply:
drawbacks of that would be, the VM's would failover together, and you will have an IO limitation as multiple VM's are now running on VHD files on the same clustered NTFS volume.
Disclaimer: I have not tested this (yet).
rgds,
Edwin.
------------------------------------
Reply:
I'd say "No." The Microsoft model has always been a "shared-nothing" model, which precludes any node from accessing a disk resource while it is being controlled by another node in the cluster.
I'd second Edwin's suggestion. You could always pre-allocate LUNs and then simply apportion a chunk to each new VM created. This of course places additional dependencies on a single LUN...with multiple VMs as part of the LUN, you would stand to lose more than a single VM system if you lost a given LUN.
With a larger cluster, you'll also likely run into limitations with using drive letters vs. the number of potential VM drives (you're capped at around 24 or 25, depending on the physical config of your host Cluster nodes). Have you looked into using passthrough disks to gauge if this will be a more flexible solution for you?
Hope this helps,
--Ryan
Ryan Sokolowski | MVP - Clustering | MCT, MCITP x3, MCTS x6, MCSE x2, CCNA, CCDA, BCFP
------------------------------------
Reply:
As long as you understand that if one of the virtual machines running on that LUN (Physical Disk) fails, the entire group will failover to another node in the cluster thus adversely impacting HA for those VMs. You can, of course, prevent this behavior by manually modifying the resource properties to 'not effect the group'. To 'properly' ensure HA, place one VM per physical disk resource.
Thanks.
Chuck
Chuck Timon Senior Escalation Engineer (SEE) Microsoft corporation
------------------------------------
Amd Dual core users, Fx 60 X2 users brand new new dual core optimizer avail
Just founnd this it may have been discovered by others already This was date August 14th 2008 available here
http://www.amd.com/us-en/Processors/TechnicalResources/0,,30_182_871_9706,00.html
Hopefully improves thous of us using the A8n 32 deluxe mb.
Liscens stas for USA usage. Sorry out side the USA. See your local Amd web site.
From the notes
Supported OS(s)
* Microsoft® Windows® XP
* Microsoft Windows Server 2003
* Microsoft Windows Vista
See the note for you gaming system users. And other issues with ATI drivers. *dont say I didnt mention this issue.
Have fun Keith
Why I Love URL Shorteners
Have you ever received an e-mail like this?
When you click on the link, it takes you to a bad web page because not all of the link was sent to the browser.
As a project manager, I receive and send tons of links via e-mail and more often than not, I would come across this problem on a regular basis. First off, the link is not bad - it's just that it's too long when it's clicked from an e-mail, the entire address is not passed to the web browser hence going to the wrong address. The right way of addressing this is to copy the entire URL and paste it on the web browser address bar.
To view the full blog entry, visit http://bit.ly/1IPTvG
better communication - add a way for moderatores to ask technical questions of the users
I know that we have to watch out for privacy on the forums, and can't make the users' email addresses viewable, even to the moderators. But if we had an option where we could have the website send mail to the users (from moderators only) asking additional clarification on an issue, it would really help. The user would have the option of ignoring that or replying. It would be all automated, so nobody human would see their personal information. It would be limited to moderators, so the users couldn't get spam.
Thanks
Anita Oakley
Reply:
I don't see the point of this.
The e-mail wouldn't have a return address (if I had I wouldn't use the function!!) so the only place the user could reply with extra information would be the forum thread itself.
If he/she's asked the question in the forum, he/she should be following the thread anyway, so posting a request for more information *in the thread* should have the same effect.
If he/she is too lazy to follow his/her own thread, why should we encourage laziness by sending him/her an e-mail ??
- Edited by Mike Walsh FIN Saturday, August 2, 2008 7:25 PM spelling
------------------------------------
Reply:
Matt Fraser, STO Forums Software Developer
------------------------------------
Reply:
Lucian Baciu, MCTS, http://studentclub.ro/lucians_weblog
------------------------------------
Reply:
So, this issue is just about users who want to be contacted by moderators but no-one else.
Oh, wait... the old forums system had a "Display Email" in the profile but the new one does not.
- Edited by Jason Kresowaty Saturday, August 16, 2008 3:46 PM Add info about Display Email not being in profile
------------------------------------
updates
Reply:
You might get your question answered if you would use english senteces.
------------------------------------
No comments:
Post a Comment