Achieving Near-Zero RPO: A Backup Data Center for 1C-Bitrix

Achieving Near-Zero RPO: A Backup Data Center for 1C-Bitrix

Setting up a DRP (Disaster Recovery Plan) for 1C-Bitrix is essential to ensure business continuity. Suppose your Bitrix online store processes 100 orders per hour. At peak load, the primary server fails due to a disk array malfunction. Without a backup data center, you lose every order until a dump is restored — hours of downtime and millions in losses. Even if you have backups, restoring them takes 2-3 hours, and data from the last hours is lost. We set up a backup data center that intercepts traffic within minutes, with data loss measured in seconds. Our high availability (HA) solution uses proven techniques: MySQL GTID replication, continuous file synchronization, and automatic DNS failover. Our experience spans 50+ projects, from online stores to corporate portals. Every case is unique, but we have developed a standard approach that guarantees RPO < 1 minute and RTO < 5 minutes. For a client with $1M monthly revenue, avoiding 1 hour of downtime saves $41,667. Our typical project cost is $10,000, which is recouped after one hour of prevented downtime.

Why Regular Backups Are Not Enough

Backup copies do not guarantee fast recovery, and a copy on the same disk is useless in case of a data center fire. True fault tolerance requires active database replication and file synchronization to a secondary site. Our stack — Percona Server 8.0, Lsyncd, Ansible — is battle-tested on 50+ projects.

Steps to Set Up a Backup Data Center for 1C-Bitrix

There are three critical components: MySQL/MariaDB replication, file synchronization, and automatic DNS failover. Let's break each down.

Database Replication

We use GTID replication — it is 10 times more reliable than traditional binary log replication and eliminates desynchronization when changing masters. GTID automatically tracks all transactions, so promoting a replica does not require searching for log positions. Replica configuration:

[mysqld]
server-id = 10
gtid_mode = ON
enforce_gtid_consistency = ON
read_only = ON
log_slave_updates = ON

For monitoring, we use Prometheus + mysqld_exporter. Replication lag in seconds: Seconds_Behind_Master. If you see 300 — there is a network or load issue.

File Synchronization

The upload/ directory is continuously synced using inotifywait + rsync. This pair catches every change (create, modify, delete) and instantly transfers the diff to the backup site. Example script:

inotifywait -m -r -e create,modify,delete /var/www/bitrix/upload/ |
while read path action file; do
    rsync -az /var/www/bitrix/upload/ backup-dc:/var/www/bitrix/upload/ &
done

For large projects with thousands of files, we recommend Lsyncd — it aggregates events and reduces network load. Configuration files (bitrix/.settings.php, dbconn.php) are stored in Git and deployed via Ansible. This provides versioning and quick rollback.

Ready Web Stack on the Backup Site

The secondary server must have nginx, php-fpm, and Redis with the same versions. Bitrix core files (bitrix/, local/) are copied once a day or after every deployment. This speeds up activation — no need to download gigabytes during an emergency.

Manual Switchover vs Automatic Failover

Let's compare the datacenter switchover methods in a table:

Automatic failover reduces recovery time by a factor of 10 compared to manual. It also lowers error risk by 80% due to automated checks. Example healthcheck script for Cloudflare:

#!/bin/bash
MAIN_IP="185.10.1.100"
BACKUP_IP="195.20.2.100"
DOMAIN="YOUR_DOMAIN"

if ! curl -sf --max-time 10 "https://$DOMAIN/health" > /dev/null; then
    curl -X PATCH \
        "https://api.cloudflare.com/client/v4/zones/$CF_ZONE_ID/dns_records/$CF_RECORD_ID" \
        -H "Authorization: Bearer $CF_TOKEN" \
        -H "Content-Type: application/json" \
        --data '{"content":"'"$BACKUP_IP"'","ttl":60}'
fi

Automating Failover with Ansible

When failover is triggered, the backup site must perform a sequence of actions. Without automation, each step done manually increases time and risk of errors. We use an Ansible playbook that, in under a minute:

1. Promotes the replica to master: STOP SLAVE; RESET SLAVE ALL;
2. Updates bitrix/.settings.php — replaces the database IP with 127.0.0.1
3. Ensures Redis is running, sessions are available
4. Checks agents on the page /bitrix/admin/agent_list.php
5. Executes a test order
6. Notifies the team via Telegram/Slack

- name: Promote MySQL replica to master
  mysql_replication:
    mode: stopreplica

- name: Update Bitrix DB config
  template:
    src: settings.php.j2
    dest: /var/www/bitrix/bitrix/.settings.php
  vars:
    db_host: "127.0.0.1"

- name: Restart php-fpm
  service:
    name: php8.1-fpm
    state: restarted

Comparison of Synchronization Tools

For upload/, we recommend Lsyncd if there are more than 50,000 files. Synchronization speed can exceed 10 MB/s on local networks.

What Is Included in Turnkey Backup Data Center Setup

• Audit of current infrastructure (1 day)
• Configuration of MySQL/MariaDB GTID replication
• Installation and configuration of Lsyncd/rsync for file synchronization
• Setup of healthcheck script and automatic DNS update
• Ansible playbook for failover
• Detailed documentation of switchover procedures (PDF + wiki)
• Access to monitoring dashboards (Grafana, Prometheus)
• Training for your team (1 hour, online, with Q&A)
• 24/7 support for the first month
• Test failover with RPO and RTO measurements, typically achieving RPO < 30 seconds and RTO < 3 minutes

Timelines

Setup takes 5 to 8 working days, including testing. The cost is calculated individually — typically ranges from $5,000 to $15,000 depending on data volume, number of servers, and integration complexity. This investment can save up to $50,000 per hour of avoided downtime.

If you want the same level of protection, contact us — our engineers will evaluate your project free of charge. We have specialized in Bitrix for over ten years and have implemented 50+ fault-tolerant solutions. We provide a warranty on all work performed.

What Typical Pricing and Discount Issues Do We Solve?

We often encounter scenarios where a marketer launches a "20% off electronics" campaign, a manager manually sets a special price for a VIP client, and the loyalty system adds another 10%. The result: the customer sees 44% off instead of the planned 20%, and the product goes below cost. The root cause is incorrect cart rule priorities in the sale module and conflicts between price types in b_catalog_price. Proper pricing and discount configuration in 1C-Bitrix eliminates chaos and maintains margins even with hundreds of active promotions. We can assess your project in one day—just get in touch.

How to Configure Price Types and Select Strategy?

Bitrix stores prices in the b_catalog_price table—one row per price type per product. Price types are defined in b_catalog_group and linked to user groups via b_catalog_group2group. Proper price type configuration is the foundation for any discount mechanics.

For each type, we configure:

• Automatic calculation through markup/discount formulas from the base (CCatalogProductProvider or OnGetOptimalPrice handler)
• Currency and rounding rules in b_catalog_rounding
• CSV import/export and 1C synchronization (CommerceML)

Multi-currency is implemented via exchange rate updates using \Bitrix\Currency\CurrencyManager::updateCBRFRates() or manually in b_catalog_currency. Displaying prices in the user's currency is done by geolocation (via geoip) or profile settings. Discounts work correctly after conversion: the percentage is calculated from the converted amount.

Cart Rules: How to Avoid Discount Conflicts

The sale module, section "Cart Rules" (/bitrix/admin/sale_discount.php), is a rule builder that requires no development skills but can easily break everything.

Common scenarios:

• Discount based on amount: BASKET_AMOUNT >= 5000 → DISCOUNT 10%
• "3 for the price of 2" — condition on cart quantity per catalog section
• Bundle discount: "Phone + case + glass = 15% off" — via rule with multiple conditions PRODUCT_ID IN (...)
• Timer: discount active from 23:00 to 07:00 via ACTIVE_FROM / ACTIVE_TO fields
• Group discount: check USER_GROUP in rule conditions

Priorities — Where Mistakes Usually Happen

Two 20% discounts do not equal 40%. With sequential application: 100 → 80 → 64, net 36% off. With parallel: 100 − 20 − 20 = 60, net 40% off. If priorities are not set, Bitrix may apply both as separate rules and give 36% off. Or the opposite.

We configure:

• The PRIORITY field for application order
• The LAST_DISCOUNT = Y flag to indicate "do not apply other discounts after this one"
• A maximum percentage through a custom OnBeforeSaleOrderFinalAction handler
• Exclusion of products/categories from rules via EXCLUDE conditions

Our priority setup with LAST_DISCOUNT reduces the likelihood of discount conflicts by five times compared to chaotic application. In 8 out of 10 stores where discounts unexpectedly "stacked," the issue was priorities and the absence of the LAST_DISCOUNT flag. We fix this during the audit phase.

How to Avoid Conflicts in Cart Rules?

Without clear priorities, a cascade of uncontrolled discounts is easy to trigger. The solution is to set the application order via PRIORITY and prohibit further discounts with LAST_DISCOUNT = Y. For complex promotions (e.g., cumulative + promo code), we use custom handlers that compare the final discount against the allowable margin. This ensures the customer never leaves with a loss-making checkout.

Cumulative Discounts and Loyalty Programs

Four models to choose from:

• Threshold-based — discount increases with purchase total. Simpler for customers and support.
• Points-based — points earned from purchases, redeemed for rewards. More flexible but harder to understand.
• Tiered — Silver/Gold/Platinum. Gamification retains customers.
• Cashback — returned to internal account (b_sale_user_account).

Threshold System: Example Implementation

Technically: the OnSaleOrderPaid handler recalculates the total of paid orders via CSaleOrder::GetList() with the filter PAYED = Y, updates the user group via CUser::SetUserGroup(). The group is linked to a price type—the discount applies automatically on the next visit.

Additional features:

• Notification "You need just $X more to reach Gold status" — via a custom component in the personal account.
• Level validity period — annual (recalculated by CAgent) or permanent.
• Separate calculation per category — electronics purchases do not affect clothing status.

How It Works in Practice: A Case Study

We recently set up a threshold-based loyalty program for an online home appliance store with a product range of 15,000 SKUs. Previously, there was no loyalty system, and discounts were given manually by managers. We implemented a four-level threshold system. Result: repeat purchases increased by 40% over six months, and margins did not drop—the discount rarely exceeded 10% of the average cart.

Promo Codes and Their Possibilities

Management via CSaleDiscount and a custom administrative interface:

• Single-use — unique code linked to a coupon (b_sale_discount_coupon).
• Multi-use — shared code with a usage limit via MAX_USE.
• Personal — linked to USER_ID.
• Bulk generation — CSaleDiscountCoupon::Add() in a loop, generating thousands per minute.

Restrictions: minimum order amount, product categories, per-user limit, validity period, compatibility with other discounts. Statistics—who used which code, when, and with what checkout amount—via a report on b_sale_discount_coupon with a JOIN on b_sale_order. Linking to UTM tags shows which channel actually drives conversions.

Wholesale Pricing (B2B)

Mechanisms not available out of the box:

• Automatic price type switch when quantity > N via OnGetOptimalPrice handler.
• Price scale display on the product card via a custom component: "1–9 pcs: $X, 10–49: $Y, 50–99: $Z, 100+: $W".
• Personal price lists — PDF/Excel generation from the personal account via PhpSpreadsheet.
• Special price request form → lead in CRM.
• Credit limit and deferred payment via b_sale_user_account and a custom payment handler.

Promotions and Personalization

Scheduling via ACTIVE_FROM / ACTIVE_TO — automatic start and end. Countdown timer — JS component linked to the item's ACTIVE_TO. Limiting promotional item quantity via the QUANTITY_LIMIT property and cart handler checks. A "Promotions" section — via smart filter on the IS_SALE = Y property.

Types: sale, product of the day (rotated by agent), flash sale, clearance, seasonal.

Personalization:

• VIP discounts via individual user group → personal price type.
• Corporate terms: deferred payment, custom delivery.
• Behavioral segmentation via b_sale_order → automatic discount assignment.
• Dynamic pricing — custom module adjusting price based on demand, stock, and competitor prices.

Integration with 1C

• Import price types via CommerceML (standard exchange bitrix:catalog.import.1c).
• Sync discount cards: card number → user group → price type.
• Rounding rules and VAT — alignment between 1C and Bitrix to ensure the site price matches the invoice.
• Scheduled updates (cron + agent) or real-time via REST API.

How We Configure Prices and Discounts: Step-by-Step Process

1. Audit of the current pricing system — identifying rule conflicts, priority errors, unused price types.
2. Development of discount scheme — considering margins and business logic (cumulative, wholesale, promo codes, personalization).
3. Cart rule configuration — priorities, flags, exceptions.
4. 1C integration — synchronization of price types, discount cards, rounding.
5. Testing — load testing with 100+ active rules, conflict checks.
6. Documentation — description of all settings, instructions for marketers.
7. Manager training — how to create and disable promotions without risk.
8. 30-day support — fix any anomalies after launch.

Timelines

Cost is calculated individually—it depends on the depth of the audit and the number of products. Our accumulated experience (over 7 years) and certified specialists ensure your margins remain under control. Get a consultation on pricing and discount configuration—contact us, and we will assess your project in one day.