Trade Math for Database Administration

Why Database Administrators Need Trade Math

A DBA who cannot size a database accurately may under-provision storage, causing outages, or over-provision it, wasting budget. The math in this lesson appears directly in day-to-day work: estimating how much disk space a new application will consume, interpreting performance data, and presenting findings to management. It also maps to exam objectives for Oracle DBA CA (performance tuning), Microsoft DP-300 (Azure SQL resource planning), DAMA CDMP (data quality metrics), and PostgreSQL CE (storage configuration).

---

Number Systems Used in Database Work

Computers store all data in binary (base 2), but DBAs also encounter octal (base 8) and hexadecimal (base 16) regularly — in IP addressing, memory dump analysis, and permission flags.

Binary (Base 2)

Digits are 0 and 1. Each position is a power of 2. To convert binary to decimal, multiply each bit by its positional power and sum:

`1011₂ = (1×8) + (0×4) + (1×2) + (1×1) = 11₁₀`

To convert decimal to binary, repeatedly divide by 2 and record remainders from bottom to top.

Hexadecimal (Base 16)

Digits are 0–9 and A–F (A=10 through F=15). Each hex digit represents exactly four binary bits (a nibble), making hex a compact shorthand for binary. IPv6 addresses, color codes, and memory addresses all use hex.

Conversion shortcut: group binary digits in fours from the right, convert each group to one hex digit.

`10110111₂ → 1011 0111 → B7₁₆`

Octal (Base 8)

Digits are 0–7. Linux/Unix file permissions (e.g., `chmod 755`) use octal. Group binary in threes from the right:

`101111₂ → 101 111 → 57₈`

---

Storage Units: SI vs. IEC Prefixes

One of the most common sources of sizing errors is confusing SI (decimal) prefixes with IEC (binary) prefixes. The IEEE and NIST both recognize the IEC standard (IEC 80000-13).

| IEC Name | IEC Symbol | Value | SI Name | SI Symbol | Value |

|---|---|---|---|---|---|

| Kibibyte | KiB | 2¹⁰ = 1,024 B | Kilobyte | KB | 10³ = 1,000 B |

| Mebibyte | MiB | 2²⁰ = 1,048,576 B | Megabyte | MB | 10⁶ B |

| Gibibyte | GiB | 2³⁰ ≈ 1.074 × 10⁹ B | Gigabyte | GB | 10⁹ B |

| Tebibyte | TiB | 2⁴⁰ ≈ 1.099 × 10¹² B | Terabyte | TB | 10¹² B |

Storage vendors typically advertise disk capacity in SI gigabytes, while operating systems often report in IEC gibibytes. This gap explains why a "500 GB" hard drive appears as roughly 465 GiB in Windows or Linux. Always clarify which convention a specification uses before sizing a database.

---

Database Sizing Formula

A standard DBA sizing estimate follows these steps:

1. Raw data size = Row count × Average row width (bytes)

2. Table overhead factor: Add roughly 20–30% for index space, transaction logs, and database overhead (use whatever factor your DBA guide or vendor documentation specifies for your platform).

3. Replication factor: Multiply by the number of copies (e.g., ×3 for a three-replica cluster).

4. Growth projection: Multiply the current-year size by `(1 + annual growth rate)^years` using compound growth.

5. Cost estimate: Multiply projected storage in GB by the cost per GB.

Always document your assumptions. Presenting assumptions alongside numbers is a core data-management professional practice emphasized in the DAMA CDMP body of knowledge.

---

Throughput Calculations

Throughput is the rate at which data moves. A common unit is MB/s (megabytes per second). To estimate how long a data load will take:

`Time (seconds) = File size (MB) ÷ Throughput (MB/s)`

For network bandwidth given in Mbps (megabits per second), convert to MB/s by dividing by 8:

`1 Gbps = 1,000 Mbps = 125 MB/s`

---

Basic Descriptive Statistics for Performance Analysis

When monitoring query response times, DBAs calculate three key statistics:

• Mean (average): Sum of all values ÷ count of values. Sensitive to outliers.
• Median: The middle value when data are sorted. More robust to outliers than the mean.
• Standard deviation (σ): Measures spread around the mean. A high standard deviation for query times signals inconsistent performance — a red flag during incident investigations.

In a spreadsheet, use `=AVERAGE()`, `=MEDIAN()`, and `=STDEV()` (sample standard deviation) or `=STDEVP()` (population). For a performance dashboard submitted in an incident report, label every column clearly and include units (e.g., milliseconds).

---

Safety in the Data Center Lab

Although database administration is primarily software work, lab sessions take place around server hardware. Follow these safe practices:

• Electrical safety: Never open powered server chassis. Data centers use high-density power; treat all rack-mounted equipment as energized unless confirmed otherwise by a qualified technician using lockout/tagout (LOTO) procedures.
• Ergonomics: Position monitors at eye level and keyboards at elbow height to prevent repetitive-strain injury during extended calculation and spreadsheet sessions.
• Data safety: Never enter real personal or financial data into practice databases. Use synthetic or anonymized data sets only.
• File hygiene: Save spreadsheet work frequently and back up to the designated network share before submitting — data loss during a lab session is not an acceptable excuse for an incomplete deliverable in a professional environment.