5.8. vLoad Balancer¶
vLoad Balancer is a service that provides a virtual dedicated load balancing device over the Server Segment. You can use the load balancing feature for communication with Virtual Machines in a Server Segment.
5.8.1. Available Features¶
You can use the following features in vLoad Balancer.
| Feature | Overview |
| Load Balancing Feature | A feature that balances the communication load for the Virtual Machine on the Server Segment. |
| Routing Feature | A feature that sets static routing to vLoad Balancer. |
| IP Address Delivery Feature | A feature that provides a Virtual IP (called “VIP” below) for communication between vLoad Balancer and vFirewall, and a feature that provides a Proxy IP for communication between vLoad Balancer and the load balancing destination server (called “real server” below). |
Note
- You can install one vLoad Balancer unit to each Server Segment.
- You can change the settings of vLoad Balancer from the Customer Portal.
5.8.2. Load Balancing Feature¶
vLoad Balancer PerformanceYou can specify the performance provided by vLoad Balancer using the vLoad Balancer values.
The performance of one vLoad Balancer resource is shown below.
| Item | Performance (maximum value) | Remarks |
| Traffic Processing Capacity | 20 Mbps | Processing capacity for transferring IP packets received into vLoad Balancer |
| Number of concurrent sessions | 20,000 | Number of TCP/UDP sessions that can be held simultaneously inside vLoad Balancer.
* Unlike vFirewall, when inbound and outbound communications occur, each one session is held.
|
| Number of Health Check Definitions | 10 | - |
| Number of Real Server Settings | 20 | - |
| Number of Server Group Settings | 20 | - |
| Number of VIP Settings | 4 | - |
| Number of routing settings | 5 | - |
You can add and reduce usable vLoad Balancer resource values, within the following range.
| Lower Limit | Upper Limit | Application Unit | |
| vLoad Balancer Resource Value | 1 | 50 (*) | 1 |
* The maximum value that can be set using the Customer Portal is 10. Please contact us separately if you would like 11 or more vLoad Balancer resources.
In order to perform load balancing, you can set load-balancing rules that specify targeted server, health check method and load-balancing method. You can set the following items for each load-balancing rule. See the User Guide for the setting method.
| Setting Name | Setting Details |
| VIP | From the VIP provided to the vLoad Balancer, specify the VIP to use for load-balancing rules. |
| Protocol | Selects the protocol of communication to be load-balanced from TCP or UDP. |
| Port | Specifies the port number of communication to be load-balanced. |
| Session Maintenance Method | Selects the method for maintaining sessions.
|
| Server Group | Specifies the server groups to which to apply these load-balancing rules.
|
| Backup Server Group | If the health check feature detects failures in all the real servers in the server group, a server group can be specified to receive distribution as backup devices (standby devices). |
| Header Addition Feature* | Specifies whether to enable or disable the feature that adds the x-forwarded-for header to HTTP communication. |
* HTTP header packet more than 4096bytes cannot be available. In Yokohama No.1 or Kansai1 or Saitama No.1 Data Center, x-forwarded-for field is inserted into only http Request header. If Customer uses vLoad Balancer in other Data Center, x-forwarded-for field is inserted into http Request and Response header if Header Addition feature is enabled in vLoad balancer which is added before the maintenance during November 4 from October 27, 2015. If vLoad balancer is added after maintenance, x-forwarded-for field is inserted into only http Request header.
Note
- You can set the load-balancing method when you add server groups, and you can also change them after that.
The health check feature detects real server failures. It sends pings or ICMP pings to the TCP port of the real server at 2-second intervals. If they fail 4 times in a row it is judged that the relevant real server is experiencing communication interruptions.
If it is determined that the real server’s communication is interrupted, the relevant real server is excluded from the load balancing destination server, and packets are no longer transferred. Instead, packets are sent to a different real server within the same server group.
After it has been determined that the real server’s communication is interrupted, it sends pings or ICMP pings to the TCP port of the real server at 30-second intervals. If the ping succeeds twice in a row, it is determined that the communication has been recovered. The real server is automatically reset into the load balancing destination server, and packet transmission resumes.
Note
- You can set the health check method from the Customer Portal.
- You can set health check methods for each server group.
- You can set the same health check method to multiple server groups.
- You can set TCP or ICMP as protocols for performing health checks. The operations are shown below.
| Item | ICMP | TCP |
|---|---|---|
| Monitoring Content | Performs ICMP Ping monitoring | Specifies the ports to be monitored and performs TCP port monitoring. |
| Health Check Intervals | 2 seconds | |
| Heath check intervals during downtime | 30 seconds | |
| Number of times before it is seen as down | 4 times | |
| Number of times before it is seen as recovered | 2 times | |
| Wait time between sending SYN and receiving ACK | - | 1 second |
5.8.4. IP Address Delivery Feature¶
VIP- VIP is a virtual IP address that is used when the load-balancing source and vLoad Balancer communicate. It is provided as an alias IP to the Server Segment side interface of vLoad Balancer.
- You can register multiple VIPs for one interface. You can set the maximum number of VIP using “VIP setting number” in vLoad Balancer resource.
- You can select VIPs from the available IP addresses in the Server Segment where the vLoad Balancer is installed. You can specify them from the Customer Portal when adding VIPs. VIPs are set as alias, active, or standby. Unspecified VIPs will be allocated automatically.
- Proxy IP is a virtual IP address that is used when the real server and vLoad Balancer communicate. It is provided as an alias IP to the Server Segment side interface of vLoad Balancer.
- You can register multiple Proxy IPs for one interface.
- You can select Proxy IPs from the available IP addresses in the Server Segment where the vLoad Balancer is installed. You can specify them from the Customer Portal when adding Proxy IPs. Proxy IPs are set as alias, active, or standby. Unspecified Proxy IPs will be allocated automatically.
- The number of Proxy IPs used differs according to the vLoad Balancer resource value that is used. When you change the vLoad Balancer resource value, Proxy IP will automatically be added or reduced by the system.
| vLoad Balancer Resource Value | Number of Proxy IP Used |
| 1 to 2 | 1 |
| 3 to 4 | 2 |
| 5 to 6 | 3 |
| 7 to 8 | 4 |
| 9 to 10 | 5 |
| 11 or more | One for every two additional vLoad Balancer resource values |
5.8.5. Important Points¶
- In order to increase the vLoad Balancer resources, available IP addresses in the Server Segment are required.
- Communication interruptions might occur when you change vLoad Balancer settings from the Customer Portal.
- When the communication is done by small number (from 1 to 4) of session for the reason of using application, there is a possibility that the throughput will be lower than maximum performance per resource. Because bandwidth is controlled by “Policing” setting, so retransmission will be happened when traffic exceeds the limit. When customer estimate or set vLoad Balancer resource, please take these into consideration. For resource estimation, refer to reference information below.
5.8.6. Reference Information¶
- The traffic result which NTTCommunications tested is shown in the following chart. These performances are not guaranteed. So please use as reference information.
- All traffic which passes vLoad Balancer is taget of bandwidth control based on resource level. So when traffic passes more than one times on one communication these are made the target of bandwidth control.Example: In case of resource level1 (20Mbps)Traffic: 15Mbps/communicationpasses twice: 15Mbps x 2 =30Mbps2 resources will be needed actually.
