ASK THE EXPERT - WIDE AREA APPLICATION SERVICES (WAAS)

Unanswered Question
Dec 7th, 2007

Welcome to the Cisco Networking Professionals Ask the Expert conversation. This is an opportunity to discuss about Cisco WAAS design, deployment, and troubleshooting, with Cisco expert Zach Seils. Zach, CCIE #7861, is a Technical Leader in the Advanced Services Data Center Networking Practice. His focus is the design, deployment, and troubleshooting of application acceleration solutions for Cisco's largest enterprise and service provider customers. He is also frequently engaged with partners and internal Cisco engineers worldwide to advise on the design, implementation, and troubleshooting of the Cisco Wide Area Application Services solution. In addition to working closely with partners and customers, he works closely with the Application Delivery Business Unit on product enhancements, testing, and application acceleration architectures. Prior to joining Cisco, Seils spent six years in various senior technical roles for a managed service provider.

Remember to use the rating system to let Zach know if you have received an adequate response.

Zach might not be able to answer each question due to the volume expected during this event. Our moderators will post many of the unanswered questions in other discussion forums shortly after the event. This event lasts through December 21, 2007. Visit this forum often to view responses to your questions and the questions of other community members.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 5 (4 ratings)
Loading.
Zach Seils Sat, 12/08/2007 - 07:32

Brad,

The NME-WAE is intended for small branch office environments ranging from 5-75 users and single digit Mbps WAN bandwidth.

Thanks,

Zach

sujitkr7cisco Sun, 12/09/2007 - 15:06

Hi,

i have two links with different ISP's and both the links (512Kbps)are terminated on seprate seprate cisco routers(2811).Currently we are useing one link and another one link is new.I want to loadsharing and redundency between them.Right now i am not using BGP (routers are configured in simple manner).Here is a my client cost matter,so i want .My presant network seneriois like this:-

Internet Ri(Isp1)-- Layer 3 switch (working here like as a simple switch)---firewall 1 (cisco ASA5510)--firewall 2(ciscoASA5510)---Lan.

on firewall-1 remote and site to site vpn configuered, one dmz.

my lan is in 172.16.1.xx series,dmz in 172.16.4.10, and after 172.16.1.30 is using for desktop ,192.168.x.x is managment ip.

Right now i have one another internet link(512kbps) from different ISP.

So please advise me how can i do that (loadbalancing with redundeny). once i was trying oer but not successfull due to IOs image (c2800nm-ipbasek9-mz.124-11.T.bin)and

ISP's AS.it can be possible through natting,and ACL. or ple give me any idea with OER or any solution.If possible ple send me also sample configuration.

If u want any info ,ple tell me .

Please reply me as soon as possible.

Thanks and regards,

sujeet

Zach Seils Mon, 12/10/2007 - 09:10

Sujeet,

This conversation is for discussion of Cisco Wide Area Application Services (WAAS). It sounds like your post would be better suited for the 'WAN, Routing and Switching' conversation. I recommend that you report your question in that forum.

Thanks,

Zach

sujitkr7cisco Mon, 12/17/2007 - 10:53

Dear Zach,

really i am affreied to this problem and i have been wondering for solution from last ten days .I respact your words and understand but right now i have to need you or your team help ,if possible.

I have two links with different ISP's and both the links (512Kbps)are terminated on seprate seprate cisco routers(2811).Currently we are useing one link and another one link is new.I want to loadsharing and redundency between them.Right now i am not using BGP (routers are configured in simple manner).Here is a my client cost matter,so i want .My presant network seneriois like this:-

Internet Ri(Isp1)-- Layer 3 switch (working here like as a simple switch)---firewall 1 (cisco ASA5510)--firewall 2(ciscoASA5510)---Lan.

on firewall-1 remote and site to site vpn configuered, one dmz.

my lan is in 172.16.1.xx series,dmz in 172.16.4.10, and after 172.16.1.30 is using for desktop ,192.168.x.x is managment ip.

Right now i have one another internet link(512kbps) from different ISP.

So please advise me how can i do that (loadbalancing with redundeny). once i was trying oer but not successfull due to IOs image (c2800nm-ipbasek9-mz.124-11.T.bin)and

ISP's AS.it can be possible through natting,and ACL. or ple give me any idea with OER or any solution.If possible ple send me also sample configuration.

Zach Seils Mon, 12/17/2007 - 11:39

Sujeet,

This conversation is for discussion of Cisco Wide Area Application Services (WAAS). It sounds like your post would be better suited for the 'WAN, Routing and Switching' conversation. I recommend that you repost your question in that forum.

Thanks,

Zach

sujitkr7cisco Mon, 12/17/2007 - 12:05

Dear Zach,

What is the minimum cost of loadbalancer through which we manage two ISP's internet link for middle size network.

Thanks, sujeet

Zach Seils Mon, 12/17/2007 - 23:24

Sujeet,

This conversation is for discussion of Cisco Wide Area Application Services (WAAS). It sounds like your post would be better suited for the 'Application Networking' forum. I recommend that you repost your question in that forum.

Thanks,

Zach

sujitkr7cisco Mon, 12/17/2007 - 12:15

Dear Zach,

What is the minimum cost of loadbalancer through which we manage two ISP's internet link for middle size network.

Thanks, sujeet

Zach Seils Mon, 12/17/2007 - 23:25

Sujeet,

This conversation is for discussion of Cisco Wide Area Application Services (WAAS). It sounds like your post would be better suited for the 'Application Networking' forum. I recommend that you repost your question in that forum.

Thanks,

Zach

gavin.mckee Mon, 12/10/2007 - 01:11

Hi Zach,

Could you discuss the application of WAS in the context of proprietary applications and real time data feeds. I work for a financial services company and am looking at minimising the footprint of applications that run over our WAN. These applications are proprietary order management systems and communicate over IP. We also have development sites worldwide, they use clearcase for version control, how does WAS help here??

Thanks

Gavin

Zach Seils Mon, 12/10/2007 - 09:04

Gavin,

WAAS can enhance throughput and lower bandwidth consumption for any application that leverages TCP as the transport protocol. The level of optimization provided depends on the specific application, and to some extent how the application is deployed.

Obviously it's difficult to gauge the type of optimization you would see with a proprietary application. That being said, we provide some utilities that allow you to replay a packet capture between two hosts, in between which you can enable WAAS. This allows you to test the effect of WAAS on an application in a non-production environment. You can find the tool, called TCP Replay, here:

http://www.cisco.com/cgi-bin/tablebuild.pl/fe_util

For clearcase, WAAS can help reduce bandwidth consumption (on all link types) and improve throughput (on high BDP links). This can improve application response times for operations that involve large volumes of data. In some cases, we have seen compression levels of 50-60%, but this can depend on the deployment.

Regards,

Zach

comnet.solutions Mon, 12/10/2007 - 03:20

Kindly let me know how should i configure my e1 channelised in cisco 2811 as my other nodes has cisco 1841

Zach Seils Mon, 12/10/2007 - 09:10

Preetam,

This conversation is for discussion of Cisco Wide Area Application Services (WAAS). It sounds like your post would be better suited for the 'WAN, Routing and Switching' conversation. I recommend that you report your question in that forum.

Thanks,

Zach

matthewevenden Mon, 12/10/2007 - 05:12

Zach

When would you recommend the use of pre-positioning? I have a customer who has a lot of static data on a file server in their brnach offices, would you recommend we preposition the data from the central DC to the remote site or rely on the CIFS accelaration?

Thanks

Matt

Zach Seils Mon, 12/10/2007 - 09:13

Matthew,

When I work with client, I usually limit PP to content such as software packages/patches. If you are dealing with a large amount of static data, you can consider running a single PP job at the beginning of the deployment, just to warm the caches.

Thanks,

Zach

matthewevenden Mon, 12/10/2007 - 09:37

Zach

Thanks for the information, just another note the customer has large pst files have you any recommendation around whether to pre-position first then rely on cache or pre-postion every day?

Thanks

Matt

Zach Seils Mon, 12/10/2007 - 11:22

I should first warn you that Microsoft doesn't recommended storing PST files on a local or remote file server. That being said, many customers do it.

In general the Outlook client doesn't read the entire PST file, but rather only a few byte ranges. The exception to this is when a PST file is searched, at which point the entire file can potentially be read (depending on the search criteria).

I would recommend an initial PP job, and then move forward relying on the on-demand caching functionality. If you find that this doesn't meet the user performance requirements, you can then consider and recurring PP job for PST files.

Zach

insccisco Mon, 12/10/2007 - 15:58

Hi,

is WIDE AREA APPLICATION SERVICES (WAAS)

a hardware piece?

insccisco Mon, 12/10/2007 - 17:05

what's its main use? and what about price or part numbers?

are they a series? meaning is there an entry level series, medium and then the most expensive and robust ones?

lastly, will they be a solution for WAN slowness. I mean, I have an office in Texas with 10+ users and they all use 6 or 7 applications which are published via citrix at the main office located in CT.

If I deploy this device, will it help in increasing response time ?

insccisco Mon, 12/10/2007 - 18:21

great link... however, sounds like more of a marketing point of view.

can I buy one of these devices and deploy it in my headquarters, where the citrix servers are located, and then expect an increase on the response time for my citrix users who are located at remote branches?

mrrussell Tue, 12/11/2007 - 06:09

Zach, a question on WAAS software upgrading.

When upgrading WAE's the software package (all 260Mbyte) is located on a server (i.e. PC - via http or ftp). Can the package be stored on the CM WAE instead and disstributed to its core/edge WAE's. This would be helpful for a Service Provider managing a Cussomers' WAE's

Zach Seils Tue, 12/11/2007 - 09:50

Mick,

Yes, this is possible. You can do the following:

1. Enable the FTP service on the CM with the command 'inetd enable ftp'.

2. Create a directory on the CM disk for storing images (ex: 'mkdir images).

3. Connect to the CM using an FTP client and upload the .bin WAAS images to the directory you have created.

4. When you define the image path in the CM, point the CM at itself (ex: at its own IP address).

Zach

iijanchan Tue, 12/11/2007 - 16:50

hi Zach,

Question regarding to migrating WAFS 3.0.9-b3 to WAAS

1. Is the roll back process available between WAFS and WAAS?

2. Will the existing cache stored on WAFS be retained after the upgrade to WAAS?

3. If an EdgeFE group is setup for load sharing and redundency, will the cache stores within each edgeFE in the edgeFE group be synchronised to minimise traffic transversing across WAN?

Regards,

Sam

Zach Seils Tue, 12/11/2007 - 23:24

Sam,

1. Rollback from WAAS 4.0 to WAFS 3.0 is not supported.

2. The existing cache space is split 50/50 between the CIFS object cache (WAFS) and DRE during the upgrade. The upgrade process will preserve as much as the CIFS cache content as possible (up to the total amount of CIFS cache space).

3. There is currently no cache synchronization between WAEs in a cluster.

Thanks,

Zach

mrrussell Tue, 12/18/2007 - 01:56

Zach, a question to clarify cache space. Is it only CIFS traffic that uses the CIFS object cache, and by CIFS does it mean any file types MS Office plus pdf, html, jpg etc as long as its served by an MS compatible file service? Assuming this is roughly true... However, I would then assume that a web server (IIS on a win 2003 server) is not using CIFS but would use only DRE cache for html and other web objects? Is one of the benefits of CIFS cache the ability for multiple sessions/clients on a branch site to share the objects in the cache, whereas DRE is dedicated to a single session?

If web services did use CIFS cache then this I guess would be similar to using web caching boxes. I'm not seeing much improvement on web services, so I guess its just using DRE (+ other optimisations).

Thanks

Mick

Zach Seils Tue, 12/18/2007 - 06:23

Mick,

The CIFS object cache is only used for objects accessed using the CIFS protocol. This can be any type of object, including the types you have mentioned. Since the cache operates at the object (i.e. file, directory. etc.) level, it allows us to perform application-specific operations including local acknowledgement, async writes, etc. This is in contrast to the DRE cache, which only stores chunks of TCP segments. Content from the DRE cache can only be accessed based on receiving a signature, or pointer, from the peer WAE.

Zach

mrrussell Tue, 12/18/2007 - 06:49

Thanks Zach. Finally just to cement my understanding. Web pages served from an web server (internet or intranet) will not benefit from CIFS acceleration, only DRE, so a seperate web cache appliance could still be beneficial.

Thanks

Mick

Zach Seils Tue, 12/18/2007 - 09:02

Mick,

You are correct. We have seen significant improvement combining the asymmetric acceleration of Cisco AVS with the symmetric optimization of Cisco WAAS. We are also continuing to develop our application-specific acceleration for various protocols.

Zach

Zach Seils Mon, 12/10/2007 - 17:51

Angel,

Wide Area Application Services is software that runs on a dedicated hardware platform called the Wide-Area Application Engine (WAE). The WAE is available in both an appliance and network module format. You can find more information about WAAS and the WAE hardware at the following location:

http://www.cisco.com/go/waas/

Thanks,

Zach

basheerpt Mon, 12/10/2007 - 23:39

Am not sure posting in the right place, please correct me if i am not.

I have a Cisco 1712 router in Branch Office connected through Point 2 point to HO. The branch operation is fine except the Pocket PC users. The pocket PC data transmission hangs randomly on its way of sending data.

I observed the issue and byepassed the Connection from the Cisco router, this way it worked fine; no hang. What could be the cause to hang the application in PPC while connecting thru the Router?

Any input is appreciated to fix this.

Thanks

Attachment: 
Zach Seils Tue, 12/11/2007 - 00:27

Mohamed,

This conversation is for discussion of Cisco Wide Area Application Services (WAAS). It sounds like your post would be better suited for the 'WAN, Routing and Switching' conversation. I recommend that you repost your question in that forum.

Thanks,

Zach

mrrussell Tue, 12/11/2007 - 06:11

Zach, a question on rollback.

If we need to rollback a version (i.e. security flaw) it would be much better if the bin file was nearer the WAE's, or can it be stored on a WAE and selected easily for a reload all from a few clicks of the WAAS CM?

Thanks

Mick

Zach Seils Tue, 12/11/2007 - 09:52

Mick,

When the WAE is upgraded, a copy of the current .bin image and device configuration are stored in an internal disk partition on the disk. When you execute the 'rollback' command, the previous software image and WAE configuration is restored from the local backup. It does not pull the previous image from across the network.

Zach

mrrussell Wed, 12/19/2007 - 05:53

Zach, I performed a restore rollback from 4.0.13 (b23) to 4.0.13 (b12) and worked as expected. However as the CM is on b23 I don't think it talks properly to the rolled back WAE. Its Online, pingable, but I cannot telnet or reach by device GUI. When upgrading WAE's one should always upgrade the CM last. However a rollback will make the WAE a lower version than the CM hence this may be the reason the CM cannot talk to the rolled back WAE ??

I'll try ftp copy of the b23 image back the WAE now.

Thanks

Mick

Zach Seils Thu, 12/20/2007 - 04:39

Mick,

Currently the CM needs to run the oldest version, so it should be upgraded last and rolled back first.

However, there shouldn't be a communication issue to the rolled back WAE. Can you execute the 'sh flash' command on the rolled back WAE and make sure the flash-based and disk-based code are the same version?

Thanks,

Zach

mrrussell Tue, 12/11/2007 - 06:12

Zach, can we expect to see the caching of MAPI messages and attchments on the local WAE soon, rather than just MAPI accelleration?

Thanks

Mick

Zach Seils Tue, 12/11/2007 - 09:54

Mick,

Content accessed using MAPI is cached today using DRE. In fact, the default action for MAPI traffic is full optimization (TFO+DRE+LZ). We are planning to add additional MAPI-specific acceleration (focused on latency mitigation) in a future release.

Zach

mrrussell Tue, 12/11/2007 - 23:27

Will this be cached e-mails, so as users on the same branch office site getting the same e-mail, can all benefit from the caching, rahter than every recipient getting the e-mail (and attachment) from the centralised Exchange server?

Thanks

Mick

Zach Seils Tue, 12/11/2007 - 23:35

Mick,

The DRE cache is a TCP segment cache. So each user will still request the email/attachment from the origin server, but the transfer across the WAN (through the optimized connection) will be highly compressed.

Zach

mrrussell Tue, 12/11/2007 - 23:53

Zach, I guess a future release will allow the sharing of attachments and maybe message bodies in cache visible to all clients using a particular edge WAE. Some Cisco competitors already do this.

Thanks

Mick

Zach Seils Wed, 12/12/2007 - 07:15

Mick,

We are focusing on improving end user response time through latency mitigation techniques. As mentioned previously, we already perform byte-level caching for MAPI traffic.

Thanks,

Zach

Actions

This Discussion