Major Internet Service Provider (ISP) Migrates to HPE NonStop for Extreme Availability with No Downtime

Copy link to clipboard
Email link
Print

Figure 1 — Original Active/Passive Architecture

Situation

A major Internet Service Provider (ISP) is providing 24×7 email, Internet, instant messaging, and other services for millions of international customers.

The ISP originally had a primary and disaster recovery farm of servers running Sybase database managers on Linux. Each farm had 16 login servers, which were geographically separated for the purpose of disaster tolerance and were interconnected by a wide area network (WAN). The ISP used Sybase data replication for Disaster Recovery between the farms (Figure 1).

Problems

The user base grew, which created a capacity strain on the Sybase systems. The databases on the individual servers were standalone and not integrated, which made querying and management slow and difficult.
Database sharding (physical partitioning of one database across multiple database instances) posed data consistency and corruption challenges.
The amount of data per database instance had grown so large that queries scanning the largest tables often timed out before completing.
The application needed more capacity to support a rapidly growing user base.

The ISP performed a proof of concept (POC) to test many of the major commercially available database engines. One of the competitors could not complete the tests. Another limped along under full load, not able to complete queries within the desired service-level agreement (SLA). Fortunately, HPE Integrity NonStop hardware with an SQL/MP database in an active/active configuration satisfactorily completed all tests.

The ISP faced a major challenge: how can it migrate from the old Linux/Sybase servers to the new NonStop architecture without impacting its customers?

Solution

- Use HPE Shadowbase Zero Downtime Migration to migrate to a Shadowbase fully active/active architecture for continuous availability (Figure 2)
- Use HPE Shadowbase Professional Services to assist with the migration
- Use bulk copy program (BCP) for the custom load (a Sybase utility)
- The ISP wrote a custom compare program

Original Architecture

Step 1: Create New HPE Shadowbase Active / Active Architecture

Step 2: Use HPE Shadowbase Change Data Capture (CDC) to Create 14 Staging Databases

Step 3: Load New Environment with Bulk Copy Program while HPE Shadowbase Continues to Queue the Changes

Step 4: Use HPE Shadowbase to Synchronize the New HPE NonStop Active/Active Environment with the Queued Changes

Step 5: Use a Custom Compare Tool to Validate the Old and New Environments

Step 6: Scale up Users in the New HPE Shadowbase Environment

Final Architecture

Figure 2 — Final Active/Active Architecture

Outcomes

Eliminates the business risk of migration by safely testing and subsequently migrating several hundred million user accounts without a service impact
Simplifies capacity management and load balancing
Optimizes for queries and transaction processing
Provides continuous application availability

HPE Shadowbase Products of Interest

HPE NonStop Shadowbase Basic Replication Software (BE441AC/QSA49V6)
HPE NonStop Shadowbase Advanced Replication LTU (BE442AL/QSA50V6)
HPE NonStop Shadowbase Compare Software (BE445AC/QSA53V6)