- CheckMates
- :
- Products
- :
- General Topics
- :
- VSX cluster and two ISP providers
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Mute
- Printer Friendly Page
Are you a member of CheckMates?
×- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
VSX cluster and two ISP providers
I’m in doubt about applying VSX cluster and Virtual Systems (VSs) in case of internet access over two ISP providers. I have two internet links from two different ISP providers and two Public DMZs subnets from every provider. The picture below shows connections between equipment and two providers.
I have a plan to change two firewalls (FW01 and FW02) with two Check Point 5600 appliances configured in Load Sharing (VSLS) VSX cluster. On the cluster two VSs will be configured, one VS for every ISP provider. On each VS OSPF routing protocol will be used for routing to the inside network, and default route on each VS for routing to the internet. Also, on each VS the following blades will be enabled: Firewall, Identity Awareness, Mobile Access, IPS, Application Control, URL Filtering, Anti-Virus, Anti-Bot, Anti-Spam, and Content Awareness. Does anyone has an experience with applying CP in this configuration. Any suggestions are welcome.
Regards,#
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I presume that your ISPs given you address space smaller than the /24.
In which case I would suggest looking into ISP Redundancy:
as a possible solution and sandwich another layer 2 switch (or VLANs) between ISPs and the external interfaces of your VSX'.
Running just two VSX units in VSLS is a questionable proposition (in my personal opinion).
Using your design will complicate things for remote clients, as it will require MEP and you will rely on external probing to route inbound traffic to DMZs.
How many other VS' are you expecting to run on this cluster?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Vladimir,
Yes, you're right, address spaces from ISPs are smaller than the /24.
I would like to have HA with two CP5600 appliances and route each public DMZ subnet through corresponding provider’s link. In case of ClusteXL HA configuration, I have to choose one provider’s link as default gateway, and will have a problem with routing the second public DMZ subnet over the second provider’s link. One solution for this configuration can be use Policy-Based Routing for the second DMZ subnet, but this solution has limitations regarding supports for some features and blades and I would not use it. So two VSs will be quite enough in this case in my opinion.
Also, you wrote that “you will rely on external probing to route inbound traffic to DMZs”. Can you give me more explanation about that.
Regards,
Vedran
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Sure.
Since in your diagram you are making distinctions between Public DMZs, these will not, by themselves be redundant.
I.e. If one of the providers will go down, external hosts will not be able to reach resources located in its corresponding Public DMZ Provider # subnet.
To attain redundancy in less than /24 public services you are looking at dynamically changing their DNS records.
This typically done by your public DNS provider service probing , something like periodic queries to designated targets on designated ports.
When the query fails, the records change to point to alternate IPs.
If you are using multiple ISPs in the fashion you are describing and are supporting P2P VPNs, you'll have to setup multiple tunnels with your peers and convince them to have their applications rely on host names, instead of hard coded IPs.
If your P2P VPN domain contains private networks, you are also will be looking for additional NAT magic and DNS trickery.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Thank you Vladimir. Your explanations were helpful to me.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Dear Vladimir,
ISP redundancy is not a possibility for this configuration, it's not supported on VSX.
You can't probe the two ISP routes, because you can't configure this.
Wolfgang
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
You are correct: VSX supported features on R75.40VS and above
But I was under the impression that the choice to use VSX vs normal cluster was yours specifically because of what you were trying to accomplish.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
And you may find this useful, if you are going to host servers with ISP redundancy in load sharing mode:
Static NAT fails for outgoing connections through gateway with ISP Redundancy in Load Sharing mode
