1.2.2 Release

[ahenning]

Note: versions are not backwards compatible

1.2.2
Is a maintenance update.
Additional datastore and traffic integrity checks.
Beta Core-Core capability. Please feedback any issues with this.
Bidirectional stats updated.
MultiSite numbering bug fixed.
To upgrade from 1.2.1 to 1.2.2 follow this guide:

1.2.1
Is available for download at http://wanos.co/wan-optimization/download/. This is a raw disk image, which can be written to HDD/USB/Compact Flash/Mounted in a virtual environment or converted to another virtual disk format. The OVA package is the recommended image for virtual environments and available at http://wanos.co/wan-optimization/download/.

Changes:
Maximum Express datastore has increased to 60GB. Min disk space required is 64GB.
Minimum memory required increased to 2 GB.
Major improvements in the deduplication engine.
Performance improvement in the compression engine.
Minor GUI improvements.
Beta UDP Optimization available.
Beta UDP Encapsulation available (firewalls and PAT).
Multiple Cores per edge supported via MultiSite config.
Service restart times have been reduced to <1 second.
Peer Discovery. This also enable keep alive and timeouts, which improves auto bypass when peers go down. Hold time is configurable in seconds.
Bidirectional Optimization. Edges now optimize Edge<->Core and Edge<->Edge.
Unlimited Edge<->Edge optimization enables very large optimized networks. E.g. its theoretically possible to have 1000 fully meshed Edge<->Edge sites.
Unless stated all Wanos versions are not backwards compatible.

[ASGH]

Great work!
I have downloaded 1.2.1 but not yet have tried it. I am wondering if all of the listed features and improvements are available for free? 
Also, do I need to expect more efficient optimization due to the improvements in deduplication and compression techniques, or it is another kind of improvement such as quality and arrangement of the code?

[ahenning]

Hi Ahmed,

Everything listed is available in Express.
Any improvements listed are measurable.

[peter]

Great job and thanks.

One thing I would like to be cleared:
Edge to Edge, Edge to Core and Core to Edge, optimization is working.
Core to Core, optimization is not yet.
If there is a core and a lot of edges, optimization is working.
Please show what network topology is when core to core optimization needed?  Would be optimization between the networks such as following:

headoffice1(core) ---- branch11(edge) ( optimization  ok)
headoffice2 (core) --- branch21(edge) ( optimization  ok )
headoffice1(core) --- headoffice2 (core) ( no optimization)

chance of work around ( would it be optimization )
put a dummy headoffice and rest of them into edge mode.
dummyhead (core) --- headoffice1(edge)
                            --- headoffice2 (edge)
                            --- branch11 (edge)
                            --- branch12 (edge)

[ahenning]

Hi Peter,

That is correct, the design was to provide a solution for the majority of small companies with a Head Office and three or four remote offices. That is really where wanos was initially aimed at. If we look at the traffic patterns of these companies, 80%-90% of the time the problem area is getting data to the remote users. Since then larger companies have started using Wanos in more complex networks and bidirectional optimization is a nice to have in case to reduce traffic in the upstream direction as well. Then in even larger and more complex networks Core-Core is also a bonus between Head Offices or a DR site.

A dummy site can be used to "engineer" a work around, but in fact a Core is not even required. All sites can be Edges and this could be say 100 fully meshed sites. There is no design requirement for Core-Core, it is merely a design option to provide better reduction. Cores are important in the network since they are configured to provide the best reduction (Within 10% of Silver-Peak in realistic cold tests). In your scenario an option would be to make one Head Office an Edge: HO1(core) -> HO2(edge). Optionally also configure all sites in MultiSite on each device to ensure the peer discovery / auto detection / fail-over continues to work on each individual site level.

A lot of work has gone into the dedup engine over the last few months and Core-Core is now undergoing some final testing before the release. The alpha results are very good and about ready for beta testing. Then if all four sites have enough resources, they can all be configured as cores. For larger networks it would make sense to use a few cores and the rest edges.


Update: Core-Core is ready for beta testing