Primary Storage

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

"Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration"

So, in a vSphere environment with two FS5200. Would volumes present themselves as a volume backed by four controllers (as in HyperSwap)?

Or would I need to to manual failover (in vSphere etc.) in case of one FS5200 completely failing ?

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Based on the fact that the storage is an enterprise system and Policy-Based HA is actually a "TechPreview" with minor limitations, it should be at least questioned in an enterprise environment.

So for me the main questions are: Will there be a way to "migrate" a HyperSwap into a "Policy-Based HA" Environment? And do you think that 8.6.0.* is actually save for production?

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi Patrik,

Please note that  Policy Based HA is not a tech preview, it is a fully released version in 8.6.1, which is a fully supported and released build of code.   Yes there are some limitations to the implementation as it is the initial launch, but additional function will be coming in each subsequent release, and for clients who are interested in a high availability solution, PBHA should be evaluated to determine if the feature set meets the requirements (and if not, what is missing and when it is expected to be supported).

WRT to 8.6.X:  Client are recommended to CD release should they want to use a feature provided by that level with the knowledge that there won't likely be any PTFs for that CD and they will need to upgrade to the next LTS if they want to get onto a different later stream (so you may end up with an extra upgrade step).  It is not a question of safe for production or not, all our releases get a full set of system test, regression test and development rigor.  We would not ship the code if we didn't expect someone to be on it.  Some clients are more risk averse and won't install code that hasn't had multiple field fixes or have slow change policies for massive estates that have to stay on the same change level, for those clients it makes more sense to run an LTS recommended level.

In terms of Migration:  In 8.6.1 you cannot migrate the hyperswap config to a PBHA and maintain your high availability, which is why for new system set ups in particular, everyone should be looking at designing in PBHA rather than hyperswap or PBR instead of GM/GMCV.  Be sure to ask about it in more detail at this week's DACH storage strategy days, there are several sessions dedicated to policy based replication (async and HA) at the event this week and at the regional UGMs in November (https://www.ibm.com/support/pages/ibm-storage-virtualize-user-group-meetings)

Cheers,

Evelyn

Based on the fact that the storage is an enterprise system and Policy-Based HA is actually a "TechPreview" with minor limitations, it should be at least questioned in an enterprise environment.

So for me the main questions are: Will there be a way to "migrate" a HyperSwap into a "Policy-Based HA" Environment? And do you think that 8.6.0.* is actually save for production?

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter

Hi Davide,

Policy-Based HA is the replacement for hyperswap and any Stretched cluster configuration. It  is available starting from  8.6.1 for FlashSystem 5200, FlashSystem 7x00, FlashSystem 9x00 and SVC.

Similarly, Policy Based Asynchronous Replication (released in the 8.5.x release stream)  is the replacement for the existing Global Mirror and Global Mirror with Change Volumes features.

Here is the announcement letter: https://www.ibm.com/docs/en/announcements/storage-virtualize-software-861

Release notes for 8.6.1: https://public.dhe.ibm.com/storage/san/sanvc/release_notes/861_releasenotes.html

Publications for the feature start here: https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

Redbook on Policy Based Replication: https://www.redbooks.ibm.com/redpapers/pdfs/redp5704.pdf

Cheers,

Hello Evelyn,

I haven't heard of this new feature, there is other material to further explore the solution? i see that is available also on SVC 8.6, so it is an alternative to Enhanced Stretched Cluster?

thanks for your contribution.

If this is a new configuration I would highly recommend looking into the newly release High Availability feature with the 9500 instead of hyperswap.

https://www.ibm.com/docs/en/flashsystem-9x00/8.6.x?topic=scenarios-policy-based-high-availability

It is much easier to configure, is based on more modern, scaling and performant technology.

Hi guys

I'm building a new hyperswap topology consisting of 2 FS9500 in the datacenters. When creating the topology I'm asked to set 2 partnership parameters: 'link bandwidth' and 'bakground copy rate'. I know that I can change them afterwards (thank god) but I'm couldn't find any usful description of what these 2 parameters actually represent.

So what exactly is meant by 'link bandwidth'? Is it the total physical available bandwith for the internode communication between the datacenters? In our case that would be 4x32Gb/s per fabric times 2 resulting in 256Gb/s in total. Or is it just a part of that bandwidth and I need the have some kind of reserve?

And what is considered 'background copy'? I think at least it should be hyperswap synchronization, but what else? What counts as 'background copy' and what not?

My question aims to not interfere with the settings for the "normal hyperswap writes", meaning I don't want to increase the latency for mirrored writes because of a bad set of parameters.

I hope some can enlighten me :)

Thanks and cheers

Peter