Definition
What is disaster recovery for Oracle ATG Web Commerce?
Disaster recovery for Oracle ATG Web Commerce is the set of systems and decisions that keep a Oracle ATG store taking orders when its primary environment fails.
A traditional plan restores the store from a backup onto rebuilt infrastructure. That takes hours or days and loses everything since the last snapshot. A live replica approach keeps a second, current copy of the store running and isolated, so failover is a switch rather than a rebuild.
The two approaches answer different questions. A backup answers "can we get the data back?". A live replica answers "can customers keep buying while we deal with this?". For a Oracle ATG store the business depends on, the second question is the one that costs money every minute it stays unanswered.
Why now
Why Oracle ATG disaster recovery changed on May 31, 2022
Vendor support is the quiet half of every disaster recovery plan. As long as a vendor ships patches, most incidents are prevented before they happen. Effective end of vendor support came on May 31, 2022, so there is no vendor patch coming for the next vulnerability. Every incident that a patch would have prevented now has to be survived instead.
The facts that matter for a Oracle ATG store today:
- Oracle ATG Web Commerce moved to sustaining support years ago (ATG 10 in 2018, ATG 9 in 2016) and has been effectively end of life since May 2022. Sustaining support means no new fixes, patches or certifications.
- Oracle's strategic commerce product is Oracle Commerce Cloud, a separate SaaS platform, not an upgrade path for ATG.
- Remaining ATG expertise sits with third-party support firms and a shrinking pool of contractors.
Failure modes
Where Oracle ATG stores actually fail
Disaster recovery planning tends to imagine fires and floods. Real Oracle ATG outages are more mundane and more frequent. The failure modes that a plan for Oracle ATG Web Commerce has to cover:
- A vulnerability in ATG, WebLogic, the JDK or Endeca with no vendor fix.
- A cluster node, Endeca MDEX or database failure in an estate nobody has rebuilt in years.
- Ransomware entering through the corporate network and encrypting the commerce estate and its backups together.
- Loss of the few people who still know how the deployment works.
- The backup itself: stored on the same host or account, encrypted alongside production, or restored with the intrusion still inside it.
The replica
Why a live replica beats restoring Oracle ATG from backup
Four things change when the store has a live, isolated copy instead of a snapshot on a shelf.
Zero recovery time, because nothing is recovered
The replica is already running. When the primary fails, customers are switched to the copy in seconds. There is no rebuild, no restore window and no scramble to find the person who knows how the store was configured.
Zero data loss, because the copy is current
The replica follows the primary continuously rather than nightly. Orders placed on the replica during an outage are captured and reconciled back when the primary returns.
Ransomware cannot follow
The replica is isolated from the primary's hosting account, network and credentials. An attacker who owns production has no path into the copy, and you never restore from a snapshot that might carry the infection with it.
Unpatched is survivable when it is not a single point of failure
Oracle ATG Web Commerce will not receive another vendor fix. The replica does not change that, but it changes the cost of the next vulnerability: the exposed primary can be taken down to contain an incident while the isolated replica keeps customers buying.
How it works
How the Jestr replica works for a Oracle ATG store
Jestr does not re-implement Oracle ATG Web Commerce. It runs a copy of your actual store, kept current and kept apart.
Jestr builds a live copy of the store
Jestr replicates the Oracle ATG store as it actually runs: the application, its data and its configuration, at the exact version and with the exact dependencies the primary uses. Nothing is re-implemented, so what customers see on the replica is the store they know.
The copy stays current and stays isolated
The replica follows the primary continuously, at a known-good state, so an update that breaks production does not break the copy. It runs on Jestr's side of the outage with no shared credentials, network or identity, so a compromise of the primary stops at the primary.
When the primary fails, you switch
One action moves customers to the replica. They browse the same catalog, sign in to the same accounts and check out. Orders are captured on the replica. The switch works even when your own network is the thing that is down.
When the primary is back, you reconcile and switch back
Orders and account changes made on the replica are reconciled into the restored primary. You choose when to switch back. Nothing was lost and no backup was restored.
What the replica captures for Oracle ATG Web Commerce
- Catalog, pricing, promotions, content and search data, so the storefront and B2B experiences work on the replica.
- Profiles, carts and the order pipeline, with orders captured during an outage and reconciled to fulfillment afterwards.
- The exact ATG, WebLogic and Endeca runtime, kept current and isolated from the primary.
- Separation from the primary's network, credentials and identity systems.
Side by side
Backup and restore vs a live Oracle ATG replica
The same outage, handled two ways.
| Dimension | Traditional backup and restore | Live Jestr replica |
|---|---|---|
| Time until customers are served again | Hours to days: provision, restore, reconfigure, test | Seconds: the replica is already running |
| Data lost | Everything since the last snapshot, often a full day | Nothing: the replica is current, and changes made during the outage are reconciled back |
| Ransomware | Backups often sit in the blast radius, and a restore can bring the infection back | The replica is isolated; you switch to it instead of restoring into the compromise |
| Unpatched platform | Every new vulnerability is permanent and the only copy is exposed | The exposed primary can be contained while the isolated replica serves customers |
| Rehearsal | Rarely tested, because a full restore is disruptive and slow | Switching to the replica is routine and can be rehearsed any time |
| Who does the work at 3am | Your team, from a runbook that may be out of date | The switch is one action; the copy was built and kept current in advance |
FAQ
Frequently asked questions
Does Jestr replace our Oracle ATG backups?
No. Keep backups for archives, audits and long-term retention. Jestr replaces the part of the plan that backups are bad at: keeping the store selling while the primary is down, without a restore window and without data loss.
Can customers actually check out on the replica during an outage?
Yes. The replica runs the real Oracle ATG store, including the cart and checkout path. Orders placed during the outage are captured and reconciled to the primary when it returns. Payment and fulfillment integrations are configured as part of onboarding.
How is ransomware kept out of the replica?
The replica shares nothing with the primary that an attacker could use: no credentials, no network path, no hosting account. Replication is one-way and validated, so an encrypted or tampered primary does not propagate. You switch to a clean copy instead of restoring into the compromise.
How fast is failover?
Seconds. The replica is already running and current, so failover is the act of pointing customers at it. There is no provisioning, restoring or reconfiguring in the critical path.
Is it safe to keep running Oracle ATG Web Commerce at all now that effective end of vendor support has passed?
Safer than it looks, if the exposure is managed. The replica does not patch the software, but it means the exposed primary is no longer the only copy. When a vulnerability is disclosed, the primary can be restricted or taken down to contain it while the isolated replica keeps customers buying. Unpatched is survivable when it is not also a single point of failure.
