Shadowbase Zero Data Loss (ZDL)

In today’s business world, consistent access to real-time, online transactional data is a competitive advantage. To realize this advantage, this data must be current and available from anywhere at all times. Otherwise, the inability to access or update, or the loss of this data carries a significant potential business cost: up to thousands of dollars per second (Figure 1 and Table 1), loss of brand reputation, regulatory fines, and other unforeseen consequences.

figure-1-average-costs-per-hour-of-downtime-across-various-industries

Figure 1 — Average Costs per Hour of Downtime Across Various Industries

(Source: Network Computing, the Meta Group, Contingency Planning Research.)

In some cases, absolutely no data loss can be tolerated:

  • Healthcare – Lost dosage records can result in patient overdose or underdose on medication.
  • Manufacturing – Car manufacturers can tolerate short production line outages, but cannot lose data regarding bolt torque settings, in case of lawsuits from accidents.
  • Electronic Funds Transfer (EFT) – Some transactions are worth millions, even if the average transaction is much lower.
  • Stock Trades – Competitive data and stock price is based on previous trades, so none can be lost.
Table 1 – Estimated Cost of Lost Data Across Various Industries
Technology ArchitectureRecovery Point Objective1Retail2Credit/Debit Card3EFT4Stock Trade5
Active/Passive and Sizzling-Hot-Takeover6~1 sec$47.5K$35.6K$688K$31.6M
Active/Active~0.5 sec$23.8K$17.8K$344K$15.8M
1Example assumes rate of 500 transactions per second
2Retail average transaction ~$95 (US Online) (Source: Monetate)
3Credit/Debit Card Average Transaction ~$71(UK) (Source: European Central Bank)
4EFT average transaction ~$1,376 (Source: Canadian Payments Association)
5Stock trade average transaction ~$63,284 (Source: London Stock Exchange)
6Asynchronous replication

The Solution

Most other data replication software uses asynchronous replication that leads to the possibility of data loss for unplanned outages. For this reason, companies must consider a zero data loss solution to protect their invaluable data. Shadowbase ZDL guarantees that all committed data changes made on one system are replicated to a backup system. Therefore, unplanned system outages never result in lost data, and the business will never face these costs or the headache of trying to recover lost data.

Shadowbase ZDL uses synchronous replication, which eliminates all possibility of data loss, as explained in these definitions:

Asynchronous Replication

  • The application’s database changes are not synchronized with the replication activity.
  • After the application makes data changes, the replication engine reads those changes, and delivers and applies them to the backup database.
  • During a failure, data changes committed on the source may not have been delivered or applied to the backup database, and this data is at risk and may be lost.

Synchronous Replication

  • Replication activity is synchronized (coordinated) with the application that is making database changes on the source.
  • Source database changes are safe-stored and/or applied to the backup database, before the application changes are committed.
  • Therefore, there cannot be updates committed on the source, which have not been saved on the backup, eliminating any possibility of data loss. No committed data is ever at risk, regardless of what happens at or to the source environment.

Shadowbase ZDL avoids data loss for active/passive architectures. Shadowbase ZDL support for active/active architectures, called Shadowbase ZDL+, is forthcoming in a future release.

Shadowbase ZDL+ Automatically Resolves Data Collisions

In active/active architectures, data collisions can occur when using asynchronous replication. By their very nature, active/active architectures run multiple application copies, with each application maintaining its own database copy. If two requests are routed to different application environments for processing, and they both act on the same account’s data, different values may result in each database. This result is referred to as a data collision. Data collisions cannot arise when using synchronous replication because the replication engine automatically resolves them. (This feature will be forthcoming in a future release, Shadowbase ZDL+.)

Table 2 compares the benefits of synchronous replication (Shadowbase ZDL and Shadowbase ZDL+) to asynchronous replication.

Table 2 – The Benefits of Synchronous Replication by Replication Mode
AttributeAsynchronous Active/PassiveSynchronous Active/PassiveAsynchronous Sizzling-Hot-TakeoverSynchronous Sizzling-Hot-TakeoverAsynchronous Active/ActiveSynchronous Active/Active
Failover FaultsYesYesNoNoNoNo
Backup UtilizedNo3No3NoNoYesYes
Data CollisionsNot possibleNot possibleNot possibleNot possiblePossibleNot possible
Data/Request PartitioningNot required2Not required2Not requiredNot requiredPossibleNot possible
Application OutageYesYesMinimal1Minimal1NoNo
Data LossYesNoneYesNoneYesNone
1All users affected, but takeover time is the same as Active/Active modes
2“Required” if run in Reciprocal Mode (Reciprocal Mode allows each system to be active for a separate data set)
3“Yes” if run Reciprocal Mode

Use Shadowbase ZDL and Shadowbase ZDL+ to significantly reduce outage costs by avoiding any data loss, eliminating data collisions, and applying active/active architecture benefits to any application. Shadowbase ZDL and Shadowbase ZDL+ are the pinnacle of replication solutions for business continuity and data integration, providing the ultimate protection level against unplanned service outage costs. Shadowbase ZDL is available now for active/passive architectures, and Shadowbase ZDL+ will be available in a future release.


Related Solutions:

Related Articles:

Related White Papers: