March 15, 2015 at 9:53 am
I am still not understanding that why the need of clustering ? and how to achieve the clustering ?
Please suggest any body.....Thanks!!
March 16, 2015 at 1:37 am
The idea behind a cluster is to have two machines, one running all your databases, and one standing by. If something happens to the first machine you get an automatic failover to the second. This increases the likelihood of your servers being online.
As for how to install it, everything is in the Books Online.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
March 16, 2015 at 5:37 am
Grant has got it right in a few words, but a bit of background might also be useful.
If you go back a few years then everything was run on physical servers and if a server failed the applications it hosted could be down for a day or so until a repair or rebuild was done. In this situation having a cluster for the most important databases made good sense. However, a cluster only ever protected the compute resource. The disk resource remained a single point of failure.
In today's world where everything is virtualised then having a cluster where all nodes are in the same data center has limited usefulness. Other techniques, such as P2P replication and Availability Groups can often be a better way to get HA, as these protect both the compute and storage resources.
A cluster where the nodes are in separate data centers can still be useful today. Where separate data centers are used the disk storage has to be replicated at the SAN or file server level, so the disk resource is no longer a single point of failure. The cluster then can provide a useful HA function, as a hardware failure at one data center allows the service to fail over and continue running at the other data center.
Although a cluster means that the passive node cannot do any useful work until a failover happens, the cost of compute resource is now so cheap this is not much of a problem. Additionally the second node of a two-node cluster does not need a SQL licence, so the cost savings on this can offset the cost of having some compute resource standing idle.
Original author: https://github.com/SQL-FineBuild/Common/wiki/ 1-click install and best practice configuration of SQL Server 2019, 2017 2016, 2014, 2012, 2008 R2, 2008 and 2005.
When I give food to the poor they call me a saint. When I ask why they are poor they call me a communist - Archbishop Hélder Câmara
March 16, 2015 at 5:53 am
mohan.bndr (3/15/2015)
I am still not understanding that why the need of clustering ? and how to achieve the clustering ?Please suggest any body.....Thanks!!
Check out my stairway series on this site at this link[/url]
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
March 16, 2015 at 11:26 am
Without clustering, a motherboard failure could mean that an application is unavailable for hours or days. With failover clusters the downtime for this event could be several minutes as the most (for applications that have services that need to be manually restarted), or less than 1 minute for applications that don't require human intervention to restart. I've seen this first-hand.
April 13, 2015 at 8:19 am
thanks
can any one tell differences between clustering and always on
April 13, 2015 at 8:57 am
Sree Divya (4/13/2015)
thankscan any one tell differences between clustering and always on
Always On is a marketing term.
The two different services are failover clustering and availability groups. Failover clustering is where you have two servers (or more) that share a disk. In the event of a server going down, the process is failed over to the other server which takes over. Availability groups are where you set up a database (or more than one) so that if it goes offline for any reason, it gets moved to another machine. Nothing is shared, but, this is a database by database set up as opposed to a server level setting.
"The credit belongs to the man who is actually in the arena, whose face is marred by dust and sweat and blood"
- Theodore Roosevelt
Author of:
SQL Server Execution Plans
SQL Server Query Performance Tuning
April 13, 2015 at 10:23 am
Sree Divya (4/13/2015)
thankscan any one tell differences between clustering and always on
Check out my HA stairway on this site starting at this link[/url]
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
April 14, 2015 at 10:33 am
In traditional failover clustering, there is one set of disks hosting the database files that are passed from one node to another, either when the failover process is manually launched or automatically if the active node has a hardware, OS, network failure. If all goes well, downtime is less than a minute and databases can be back online and available to users by the time you're logged on to take a look.
This protects you against a problem with the server (or the network connection to the active server), but not against the loss of disk or issue with the SAN.
In Always-On, there are two sets of disks and two copies of the user databases which are kept synchronized. In addition to protection against server-based failures, you are also protected against the loss of a data disk, or SAN issues if the two servers are connected to different SANs.
April 15, 2015 at 3:50 am
A couple things that I would add to what you might be pulling out of all of the rest of the comments I have read to far:
1. Clustering is still the only game in town for HA for SQL Server for the following reasons
2. Clustering will monitor and fail-over for any failed resource owned by the cluster group (DNS, IP, Service, Physical Disk); virtual technologies do not monitor for failed resources like this, they only move to a different physical host when the current host fails for anything, but the smallest servers. MS technologies are a slight exception.
3. Instances of SQL Server can run on the other nodes in the cluster giving HA to all instances while not having physical servers unused.
4. Traditional failover clustering would be used in most cases if you are just using Clustering for HA, but Alwayson comes into play if you have a limited budget for hardware (no san) and you can afford to pay for licensing all servers in the AlwaysOn cluster. In Traditional Failover clustering, passive nodes are not licensed. In AlwaysOn clusters all servers are licensed.
5. Clustering doesn't have to protect against (and shouldn't) protect against a failed disk on the SAN. HA for the SAN, like for the Network should also be in place in addition to the cluster's ability to fail out of either into a completely different SAN and network.
*complete side note: Clustering is required in order to achieve HA while in the cloud. That is heresy for what a lot of people think, but I assure you, it is the case.
April 15, 2015 at 7:14 am
Keith Hays (4/15/2015)
but Alwayson comes into play if you have a limited budget for hardware (no san).
If you have a limited budget and cant afford SAN then you'll likely not be able to afford the SQL Server enterprise licences.
Incidentally, A SAN cost is usually dispersed company wide and so doesn't specifically apply to one business area.
Keith Hays (4/15/2015)
In AlwaysOn clusters all servers are licensed.
There is provisioning in AlwaysOn for unlicenced passive nodes as long as they are truly passive. The stipulation is that for every licenced node you may have one passive standby, this means no readable secondary or backup offload on that node. Check the SQL Server licensing guide for more info
SQL Server 2014 Licensing Guide
Failover BasicsFor each server licensed with SQL Server 2014 and covered by active SA, customers can run up to the same number of passive failover instances in a separate, on-premises OSE to support failover events. A passive SQL Server instance is one that is not serving SQL Server data to clients or running active SQL Server workloads. The passive failover instances can run on a separate server. These may only be used to synchronize with the primary server and otherwise maintain the passive database instance in a warm standby state in order to minimize downtime due to hardware or software failure.
? The secondary server used for failover support does not need to be separately licensed for SQL Server as long as it is truly passive, and the primary SQL Server is covered with active SA. If it is serving data, such as reports to clients running active SQL Server workloads, or performing any “work”, such as additional backups being made from secondary servers, then it must be licensed for SQL Server.
SQL server 2014 Licensing guide
AlwaysOn Availability GroupsEnhanced for SQL Server 2014 Enterprise Edition, AlwaysOn Availability Groups enable customers to configure multiple databases that will failover as a unit, with support for up to eight active secondary servers and two synchronous secondary servers. The ability to use secondary servers for more than just passive failover support can improve the performance of primary, reporting and backup workloads due to better balancing of workloads across instances, helping to provide better return on hardware investment.
Note: When secondary servers are actively used to support these additional workload scenarios—that is, when the servers used for failover purposes are no longer truly passive—they must be fully licensed accordingly.
-----------------------------------------------------------------------------------------------------------
"Ya can't make an omelette without breaking just a few eggs" 😉
April 16, 2015 at 3:05 am
Good Info, its good to have specifics especially with licensing.
Some people may have skipped over the "covered by active SA" line. Software Assurance is not a cheap line item especially when adding it to the already expensive (relative to the old days) SQL Server licensing.
Viewing 12 posts - 1 through 11 (of 11 total)
You must be logged in to reply to this topic. Login to reply