Show Posts

This section allows you to view all posts made by this member. Note that you can only see posts made in areas you currently have access to.


Topics - ahenning

Pages: 1 [2] 3
16
Troubleshooting / FAQ: Peer Down
« on: March 18, 2015, 02:35:18 PM »
By default wanos is in bridge/switch mode. This means traffic flows through wanos transparently as if it was a single cable.

In a simple point to point deployment, no configuration is essential to get peers online. No traffic policies, No multisite configs. Not even IP Addresses. Simple plug and play.

Requirement to get peers online:
1) Cabled correctly: wan0 needs to connect to the wan router and lan0 needs to connect to the lan switch. Check the logs, check the macs, switch port roles if required.
2) Traffic: Peer will detect each other automatically once TCP traffic flows. When traffic stops, peers move into idle mode to make failover seamless.
3) Flow: Traffic needs to flow through both devices in both directions. Check interface bandwidth graphs to ensure traffic flows in and out.
4) Firewalls: Some firewalls strip TCP Options. If this is suspected, enable UDP encap on all ends.




17
Configuration / Router Mode
« on: March 17, 2015, 01:26:09 PM »
This post may be outdated. Please refer to the Router Mode



Support:
The Router mode feature can be enabled and configure in Express, but support is only available for Plus since routing, policy based routing, gateways and high availability can get complicated and unique to each case.

Overview:
In router mode, only the wan0 interface is used. In this mode Wanos supports two or more ip addresses on the same interface. Primary wan0, Secondary virtual lan0. Both addresses bind to the wan0 interface.

Traffic from the wan is directed to the standard wan0 ip address.
Traffic from the lan is directed to the virtual lan0 ip address.

All sites can be in router mode or a mix of router mode and bridge mode.


Typical Configuration:

This address may overlap with wan0. In most cases, this is all that needs to be configured on Wanos. Other than this setting, the network needs to be configured to direct the traffic to these two addresses. Use routing, policy based routing or default gateways to direct traffic to the correct addresses.

To verify that the virtual lan0 is active, ping the address from the network e.g. from server or router. Remember to enable promiscuous / mac spoofing or equivalent.

Stop here for the initial configuration. "VIRT_LAN0" is the only Wanos config required.



Advance setup with multiple subnets:
Optional* Multiple secondary lan0 networks are also supported, if the workstations will direct their default gateways directly to the virtual gateways configured on Wanos. To configure these multiple additional virtual lan0 networks, these extra networks can be added to /tce/etc/wanos/virtual_lan.
e.g:
10.1.1.1/24
10.2.2.1/24
10.3.3.1/24
All these networks will be available as secondary virtual ip addresses to be used as gateways for the workstations or servers.





Examples:
In the diagram the Wan Accelerator is deployed out of path at the main site. Traffic from the wan is directed to the wan0 address 10.0.0.2. Traffic from the lan that needs to be optimized is directed to the virtual lan address 10.0.0.1




Example 2:
In the following diagram the data center wan accelerator primary address shares a segment with the gateway router. Traffic from the wan is directed to the wan0 address 10.0.0.2. Devices on the lan have their default gateways set to be the wan accelerator virtual lan addresses 10.1.1.1 and 10.2.2.1. High availability is possible by configuring secondary routes.



Verification
While pinging a host at the remote site. tcpdump should produce the following output on wan0:
sudo tcpdump -i wan0 host x.x.x.x



Both the icmp echo and echo replies are seen in the output. Noting only echos and no replies, or only replies and no echos indicates that routing or default gateway or PBR needs to be reviewed.

18
General Discussion / Version 2.0.4
« on: March 15, 2015, 10:15:55 PM »
This is a minor update and compatible with v2.0.3.

Peerstatus missing info bug fixed.
Packet loss recovery status added to peerstatus
Minor gui updates: e.g. Alert footer added to remind users to apply policy changes.
UDP Encap src port changed to 4050.

Note: Express only supports the default single site configuration. MultiSite is available in plus.

Update process



Complete image downloads


19
General Discussion / Version 2.0.3
« on: March 09, 2015, 11:37:23 PM »
New:
Layer 3 Routed mode
Out of path via Routing
Out of path via PBR
Single Ethernet deployment
Simple gateway router for multiple lan subnets

Improvements:
QoS improved to not influence Packetloss Recovery and Packet Order Correction and will help some scenarios to keep the datastores better in sync.

Minor GUI's e.g:
CPU graphs now graphs each CPU thread for a more detailed view of thread utilization.

Maybes to come: Completed
Enable UDP Tunnel mode
Semi transparent Web Cache
KVM Documentation
Amazon EC2 AWS image
CentOSUbuntu image


UPDATED: Wanos Update

20
Configuration / Packet Loss Recovery Config
« on: March 03, 2015, 12:08:56 AM »
Note: Eventually this forum post will be removed, since it is a duplicate of the Packet Loss Recovery.

Note: Packet Loss Recovery constitutes three features. Fast Retransmit Replay, Fast Retransmit Request, Forward Error Correction

From v.4 onward PLR is auto configured and FEC is optional.

With 2.0.2 packet loss recovery (PLR) has been made very dynamic to changing network conditions. This also provides Packet Order Correction.
This setting is optional. The default mode is used if the peer ip is not configured.

PLR and POC are simply enabled in the MultiSite tab by specifying the remote wanos IP:







Caveats:
In the first example the remote peer ip is in the same subnet as the site subnets and no further config is needed. In the second example, the remote peer ip is in a different subnet and it is added to the site to link the two.

Requirements:
Very specific traffic policies are required on each end for the remote site. Bypass any subnets not being optimized e.g. un-optimized remote sites and WAN devices. Wanos will automatically disable the feature per peer if it detects that packets cannot be recovered e.g. destined to the router. Hence its best to have specific source-destination policies.

The default gateway needs to be on the wan0 side, else false positive peer detected on lan0 messages could be logged.

Expected Log sequence:
rsp_peer0 - State set to CLOSED
rsp_peer0 - State set to SYN_SENT
rsp_peer0 - State set to SYN_RCVD
rsp_peer0 - State set to OPEN

21
Hardware / Changing wan0 or lan0 NIC pairs
« on: February 14, 2015, 12:36:23 PM »
FAQ: How to change the wan0 and lan0 to bind on a different interface with NIC bypass:

If the bypass pair is on eth2 and eth3 interfaces:

Edit the /tce/etc/mactab file.
Change wan0 to eth2 MAC address
Change lan0 to eth3 MAC address
Reboot

22
General Discussion / Version 2.0.2
« on: January 31, 2015, 02:19:17 PM »
2.0.1 is ready for production.

Please leave feedback if you find any issues so we can investigate and if needed fix asap. Note: No backward compatibility.

Packet Loss Recovery and Packet Order Correction features are available for very early testing.

Download Link for Wanos

Wanos Update
UPDATED: Link to Wanos Update

2.0.2
It has a major throughput performance boost on recent CPU/RAM hardware. Depending on hardware configuration throughput increase is expected to be 20-130%. On one specific test system throughput increased 2.3X.

Additional safety checks for packet loss recovery and packet order correction. These features are very robust and dynamic based on network conditions. Cold compression ratio increase significantly with these enabled (will get the specific figures). Link to guide -> UPDATED: Packet Loss Recovery

Some reported missing peer status info in the Peerstatus tab at random times.

MultiSite calculated required Ram incorrectly.

Added initial Health Status/Check tab.

SSH disabled/enabled via gui settings

Datastore drive configurable via GUI. (Still requires valid partition to exist). After updating to 2.0.2 double check / reconfigure datastore drive.

As requested the policy configuration has been made easier to edit existing rules.

23
General Discussion / Thanks to all the users!
« on: January 13, 2015, 12:51:45 PM »
To the thousands of users that downloaded Wanos in 2014, trialed, deployed in production and most of all those that provided valuable feedback on how to improve the product, a very BIG Thank You!

In 2015 we have a full program lined up and hope that we can reach even more networks.

Thanks!

24
Hardware / Wanos on Expand 4930
« on: January 10, 2015, 10:35:23 PM »
Note: Skip the hassles of flashing and upgrading another appliance and simply get the Wanos Wan Optimization Appliance .


Since quite a few users are using the Expand Networks 4930 appliance for their wanos deployments, here is a short how to guide:

Pros:
There are quite of few of them around and they have little to no resell value. Hence they make good low cost wanop devices. The Peribit or Juniper WXC 250 and 500 would probably also work.
The Expand 4930 hardware is actually top quality and was probably best in wanop class at the time. The hardware seems a bit over-engineered and it might have worked out better for Expand Networks if they went leaner and added more RAM.
The bypass NIC has two modes, changing jumper JP3 makes the NIC bypass compatible with Wanos.
Dual psu if required.
Boot from HDD, Compact Flash or USB.

Cons:
Single CPU core, which means this device should probably be used on low speed links <4Mbps. It peaks at about 100Mbps on low and 20 Mbps on high, but due to the single core, its not good at those rates.
1GB Ram.

Recommendations in images:

Move the jumper (jp3) one pin up to make the bypass work with wanos:
Default jp3 setting:


When moved one pin up:


Howto upgrade the RAM to 2GB (DDR2 notebook):


While the box is open, write the wanos image with an external usb drive connector like this:




4930 Chassis:

4930 Internal:

25
Hardware / Hardware Sizing
« on: January 10, 2015, 09:43:44 PM »
Frequently asked question: What Hardware Specs are required?

Here are some guide lines:

Disk
If peak performance is important use SSD drives.
When using mechanical disks, higher RPM drivers and larger disk caches perform better. Also consider RAID.
When going the RAID route, avoid raid controllers that require a proprietary driver and semi-software raid.

CPU
In a MultiSite environment, multi-core CPU's perform better. 4+ CPU threads recommended.
In a high speed point to point scenario, a CPU with higher single-thread ratings will perform better. 4 CPU threads recommended.

RAM
Appliances in 'High' require about 1 GB for each peer.
Appliances in 'Low' require 1 GB for each 'high' peer and 0 GB for other 'Low' peers.

Case Study 1:
Head Office with 12x 2 Mbps remote branches in a hub and spoke scenario. Recommended hardware for the remote sites is the Wanos 200 or equivalent. For the Head Office Xeon E3-1230, 16 GB RAM, 256 GB SSD.

Case Study 2:
A network with 6 fully meshes 10 Mbps sites all configured in 'High'. Recommended hardware for each site Xeon E3-1230, 8 GB RAM, 256 GB SSD. For example : HP DL320e G8.

26
FAQ / RDP Optimization
« on: December 10, 2014, 12:45:43 PM »
How to remove RDP (Remote Desktop Protocol) from the default Interactive bypass group?

Edit: /tce/etc/wanos/groupobjects and remove the port 3389 from the Interactive bypass list.

Before:
Quote
.., 2727, 3389, 5060, ...

After
Quote
.., 2727, 5060, ...

Remember to also disable RDP encryption and compression.

Web tool to be added for editing the groups/labels

Note: RDP in the default interactive class receives a high priority. Removing the port from the default group could potentially have a negative impact on RDP.

27
Installation / VMware Workstation unrecoverable error: (vcpu-0)
« on: December 05, 2014, 03:26:54 PM »
Question posted elsewhere:

Quote
today i set up a wanos system on vmware workstation 10.4, i finished all configration based on the manual but when i run the virtual machine, a error occurred:
 
Msg_Post: Error
[msg.log.error.unrecoverable] VMware Workstation unrecoverable error: (vcpu-0)
NOT_IMPLEMENTED d:/build/ob/bora-1744117/bora/lib/disklib/descriptor.c:3148

28
Hardware / Hardware Sizing
« on: November 19, 2014, 12:52:57 PM »
Frequently asked question: What Hardware Specs are required?

The Performance Guide is a good hardware sizing guide.

Here are some guide lines:

Disk
  • If peak performance is important use SSD drives.
  • When using mechanical disks, higher RPM drivers and larger disk caches perform better. Also consider RAID.
  • When going the RAID route, avoid raid controllers that require a proprietary driver and semi-software raid.

CPU
  • In a MultiSite environment, multi-core CPU's perform better. 4+ CPU threads recommended.
  • In a high speed point to point scenario, a CPU with higher single-thread ratings will perform better. 4 CPU threads recommended.

RAM
  • Appliances in 'High' require about 1 GB for each peer.
  • Appliances in 'Low' require 1 GB for each 'high' peer and 0 GB for other 'Low' peers.

Case Study 1:
Head Office with 12x 2 Mbps remote branches in a hub and spoke scenario. Recommended hardware for the remote sites is the Wanos 200 or equivalent. For the Head Office Xeon E3-1230, 16 GB RAM, 256 GB SSD.

Case Study 2:
A network with 6 fully meshes 10 Mbps sites all configured in 'High'. Recommended hardware for each site Xeon E3-1230, 8 GB RAM, 256 GB SSD. For example : HP DL320e G8.

29
Troubleshooting / Insufficient disk space
« on: October 17, 2014, 10:38:59 AM »
When alerts are logged and errors are displayed during multisite configuration if disk space is not sufficient.

To add an additional partition:
sudo su
/tce/etc/wanos/wanos-fdisk sda
reboot
mkfs.ext3 /dev/sda2
reboot
set the datastore to sda2 in the gui

This is based on the assumption that the only drive installed is sda and that is where the un-partitioned free space is located. If a second drive was added it would normally be sdb. Double check the output of the commands.


30
General Discussion / Very high latency link
« on: September 27, 2014, 06:22:53 PM »
Satellite links normally have latency in the +-500ms range, but in this test the latency is double at 1000ms. WAN link size is 1 Mbps.

http://www.youtube.com/watch?v=kaJQBTTjJE0

Pages: 1 [2] 3