You can configure and manage routing for isolation segments in TAS for VMs as described in this topic. You can also deploy a set of Gorouters for each isolation segment to handle requests for apps within the segment.
For more information about how isolation segments work, see Isolation Segments in TAS for VMs Security. For more information about creating isolation segments, see Installing Isolation Segment.
The instructions in this topic assume you are using Google Cloud Platform (GCP). The procedures can differ on other IaaSes, but the concepts must be transferable.
Isolation segments isolate the compute resources for one group of apps from another. However, these apps still share the same network resources. Requests for apps on all isolation segments, as well as for system components, transit the same load balancers, TAS for VMs Gorouters, and TAS for VMs TCP Routers.
When you use isolation segments, TAS for VMs designates its Diego Cells as belonging to an isolation segment called shared
. This isolation segment is the default isolation segment assigned to every org and space. This can be overwritten by assigning an explicit default for an organization. For more information about creating isolation segments, see Installing Isolation Segment.
The following illustration shows isolation segments sharing the same network resources:
The two isolation segments each contain a single Diego Cell. These isolation segments use the same Gorouter, TCP Router, and load balancers.
Operators who want to prevent all isolation segments and system components from using the same network resources can deploy an additional set of Gorouters for each isolation segment:
The two isolation segments each use separate Gorouters and Diego Cells. However, the use of an isolated TCP router in this scenario is not supported.
Use cases include:
Requests for apps in an isolation segment must not share networking resources with requests for other apps.
The TAS for VMs management plane can only be accessible from a private network. As multiple IaaS load balancers cannot typically share the same pool of back ends, such as TAS for VMs Gorouters, each load balancer requires an additional deployment of Gorouters.
Create a network or subnet for each isolation segment on your infrastructure. For example, an operator who wants one isolation segment separated from their TAS for VMs Diego Cells could create one network named sample-network
with two subnets named sample-subnet-tas
and sample-subnet-is1
.
The following diagram describes the network topology:
IaaS network: sample-network
|
|_____ IaaS subnet: sample-subnet-tas
|
|_____ IaaS subnet: sample-subnet-is1
Subnets do not generally span IaaS availability zones (AZs), so the same operator with two AZs needs four subnets.
IaaS network: sample-network
|
|_____ IaaS subnet: sample-subnet-tas-az1
|
|_____ IaaS subnet: sample-subnet-tas-az2
|
|_____ IaaS subnet: sample-subnet-is1-az1
|
|_____ IaaS subnet: sample-subnet-is1-az2
For more information about networks and subnets in GCP, see Using Networks and Firewalls in the GCP documentation.
To configure the subnets with BOSH, use BOSH cloud config subnets. Each subnet in the IaaS must correspond to a BOSH subnet that is labeled with the correct isolation segment. For more information, see Usage in the BOSH documentation.
Go to the Assign AZs and Networks pane of the Isolation Segment tile to assign your isolation segment to the network you created in Step 1. For more information, see Step 3: Configure the Tile in Installing Isolation Segment.
Go to the Resource Config pane of the Isolation Segment tile and use the dropdown to set your Router instances to a number greater than zero. For more information, see Step 3: Configure the Tile in Installing Isolation Segment.
If your IaaS supports it, go to the Resource Config pane of the Isolation Segment tile and enter the name of your load balancer under Load Balancers. For more information, see Step 3: Configure the Tile in Installing Isolation Segment. If your IaaS does not support this configuration, you must create static IP addresses and assign them to your load balancer out of band.
Create a separate domain name for each Gorouter instance group, and configure DNS to resolve these domain names to a load balancer that routes requests to the matching Gorouters.
You must configure your load balancers to forward requests for a given domain to one Gorouter instance group only.
As Gorouter instance groups might be responsible for separate isolation segments, and an app might be deployed to only one isolation segment, requests can only reach a Gorouter that has access to the apps for that domain name. Load balancing requests for a domain across more than Gorouter instance group can result in request failures unless all the Gorouter instance groups have access to the isolation segments where apps for that domain are deployed.
It is a common requirement for apps on separate isolation segments to be accessible at domain names that share a domain. For example, private-domain.com
. To achieve this configuration while also obeying the guideline for forwarding requests for a domain to only one Gorouter instance group, create a new TAS for VMs domain for a needed subdomain, such as *.foo.private-domain.com.
The diagrams illustrate a topology with separate load balancers, but you could also use one load balancer with multiple interfaces.
In this configuration:
Requests for system domain *.cf-system.com
and the shared domain *.shared-apps.com
are forwarded to the Gorouters for the TAS for VMs Diego Cells.
Requests for private domain *.foo.private-domain.com
are forwarded to the Gorouters for IS1. Requests for private domain *.private-domain.com
are forwarded to the Gorouters for IS2.
The example has three isolation segments and three Gorouters.
The three isolation segments each use separate Gorouters and load balancers.
Configure firewall rules to allow for necessary ingress and egress traffic for isolation segments and TAS for VMs Diego Cells. Assuming a default deny-all rule, properly configuring firewall rules prevents a request with a spoofed Host header from being forwarded by a Gorouter to an app in a different isolation segment.
Firewall rules are specific to each IaaS, so the exact definition of Source
and Destination
depends on the IaaS. For example: * On GCP, a Source
is a subnet and a Destination
is a tag. * On AWS, both Source
and Destination
are security groups.
To configure firewall rules for isolation segment traffic:
Configure the firewall rules in the table below:
For information about the processes that use these ports and their corresponding manifest properties, see Port Reference Table.
Rule Name | Source | Allowed Protocols/Ports | Destination | Reason |
---|---|---|---|---|
tas-to-bosh |
TAS for VMs Diego Cells | tcp:4222, 25250, 25777 |
BOSH Director | BOSH Agent on VMs in the TAS for VMs Diego Cells to reach BOSH Director |
tas-internal |
TAS for VMs Diego Cells | tcp:any, udp:any, icmp:any |
TAS for VMs Diego Cells | VMs within the TAS for VMs Diego Cells to reach one another |
tas-to-is1 |
TAS for VMs Diego Cells | tcp:1801, 8853, 53035 |
Isolation segment | Diego BBS in TAS for VMs Diego Cells to reach Diego Cells in isolation segment |
is1-to-bosh |
Isolation segment | tcp:4222, 25250, 25777 |
BOSH Director | BOSH Agent on VMs in isolation segment to reach BOSH Director |
is1-internal |
Isolation segment | tcp:all, udp:all, icmp:all |
Isolation segment | VMs within isolation segment to reach one another |
is1-to-tas |
Isolation segment | tcp:3000, 3001, 4003, 4103, 4222, 4224, 4443, 6067, 8080, 8082, 8083, 8443, 8447, 8844, 8853, 8889, 8891, 9000, 9022, 9023, 9090, 9091 |
TAS for VMs Diego Cells | Diego Cells in isolation segment to reach Diego BBS, Diego Auctioneer, and CredHub in TAS for VMs. Loggregator Agent to reach Doppler. Syslog Agent to reach Log Cache Syslog Server. Gorouters to reach NATS, UAA, and Routing API. Metrics Discovery Registrar to reach NATS. |
(Optional) Configure the firewall rules in the following table:
Rule Name | Source | Allowed Protocols/Ports | Destination | Reason |
---|---|---|---|---|
jumpbox-to-is1 |
Jumpbox VM | tcp:22 |
Isolation segment | Jumpbox VMs to reach isolation segment through SSH or BOSH SSH. |
diego-cell-egress |
Diego Cell VM on isolation segment | tcp:any, udp:any |
Internet | If Diego Cells must download buildpacks to stage apps, allow egress traffic from all Diego Cell VMs on isolation segments to reach the Internet. |
Additional firewall rules might be necessary to allow logs to egress to application syslog drains. Connections for syslog drains are initiated both from Diego Cells and Routers.
For more information about ports used by agents to communicate with BOSH, see the bosh-deployment repository on GitHub.
For more information about networks and firewall rules for GCP, see Using Subnetworks in the GCP documentation.
To understand which protocols and ports map to which processes and manifest properties for the preceding rules, see the following table:
Protocol | Port | Process | Manifest Property |
---|---|---|---|
tcp |
1801 |
Diego Rep | diego.rep.listen_addr_securable |
tcp |
3000 |
Routing API | routing_api.port |
tcp |
3001 |
Routing API | routing_api.mtls_port |
tcp |
4003 |
VXLAN Policy Agent | cf_networking.policy_server.internal_listen_port |
tcp |
4103 |
Silk Controller | cf_networking.silk_controller.listen_port |
tcp |
4222 |
NATS | nats.nats.port |
tcp |
4224 |
NATS | nats-tls.nats.port |
tcp |
4443 |
CAPI Blobstore Port - HTTPS | capi.blobstore.tls.port |
tcp |
6067 |
Log Cache Syslog Server | log-cache.log-cache-syslog-server.syslog_port |
tcp |
8080 |
CAPI Blobstore Port - HTTP, Diego file server - HTTP | capi.blobstore.port, diego.file_server.listen_addr |
tcp |
8082 |
Doppler gRPC, Reverse Log Proxy Gateway | loggregator.doppler.grpc_port, loggregator.reverse_log_proxy.egress.port |
tcp |
8083 |
Log Cache cf-auth-proxy | log-cache.log-cache-cf-auth-proxy.proxy_port |
tcp |
8084 |
Diego file server - HTTP | diego.file_server.listen_addr |
tcp |
8443 |
UAA | uaa.uaa.ssl.port |
tcp |
8447 |
Diego file server - HTTPS | diego.file_server.https_listen_addr |
tcp |
8844 |
CredHub | credhub.credhub.port |
tcp |
8853 |
BOSH DNS health | health.server.port from bosh-dns-release |
tcp |
8889 |
Diego BBS | diego.rep.bbs.api_location |
tcp |
8891 |
Diego Database (Locket) | diego.locket.listen_addr |
tcp |
9000 |
Loggregator Syslog Binding Cache | loggr-syslog-binding-cache.external_port |
tcp |
9022 |
Cloud Controller Stager | capi.stager.cc.external_port |
tcp |
9023 |
Cloud Controller TPS | capi.tps.cc.external_port |
tcp |
9090 |
Cloud Controller Uploader | capi.cc_uploader.http_port |
tcp |
9091 |
Cloud Controller Uploader | capi.cc_uploader.https_port |
tcp |
53035 |
System Metrics Scraper | system-metrics-scraper.loggr-system-metric-scraper.scrape_port |
tcp |
25250 |
BOSH Blobstore | bosh.blobstore.port |
tcp |
25777 |
BOSH Registry | bosh.registry.port |
For more information, see Understanding backend services in the GCP documentation and the BOSH Google CPI Release repository on GitHub.
You can configure Gorouter sharding for isolation segments depending on your use case:
Use Case | Description | How to Configure |
---|---|---|
Securing apps that run in an isolation segment | To provide security guarantees in addition to the firewall rules described above, you can configure sharding of the Gorouter's routing table, resulting in a Gorouter dedicated to an isolation segment that only recognizes routes for apps in the same isolation segment. |
|
Deploying additional Gorouters for TAS for VMs | The flexibility of the configuration also supports deployment of a Gorouter that excludes all isolation segments. |
|
Note For compute isolation only, you can select Accept all under Requests to isolation segments in the Networking pane of the TAS for VMs tile. This is the default setting, which does not require any additional Gorouters for the Isolation Segment tile.
For metrics emitted by the Gorouter, metrics can be distinguished by the name of the job. For example, this line is a metric emitted on uptime
:
origin:"gorouter" eventType:ValueMetric timestamp:1491338040750977602 deployment:"superman.cf-app.com" job:"router_is1" index:"9a4b639c-8f0e-4b2b-b332-4161ee4646e6" ip:"10.0.16.23" valueMetric:<name:"uptime" value:118 unit:"seconds" >