Modern enterprise Software-as-a-Service (SaaS) platforms have reached a point where the sheer volume of transactional data often outpaces the efficiency of traditional relational storage models. As these platforms grow, managing data effectively has become essential for maintaining performance and controlling operational costs.
Over time, large volumes of historical case data, attachments, and audit trails burden system resources, resulting in increased latency, elevated memory consumption, and escalating subscription costs. Consequently, the implementation of a data lifecycle management strategy is no longer a peripheral IT concern but a central pillar of enterprise architecture.
The cloud environment operates on a two-tier storage model designed to balance fast system performance with long-term data retention needs. This model separates storage into Cloud Data Storage (CDS) and Cloud File Storage (CFS), with each serving a different purpose in managing data thought out its lifecycle.
Cloud Data Storage represents the “hot” tier in modern SaaS platforms. It is typically built on a relational database and stores active business data, application rules, and live process or case information.
This tier is designed for high-frequency read and write operations, supporting the real-time execution of complex workflows and AI-powered decisioning engines. Because the database must maintain strict ACID (Atomicity, Consistency, Isolation, Durability) compliance and support relational queries across hundreds of tables, it utilizes high performance solid-state drives (SSDs) and dedicated compute resources.
Cloud File Storage serves as the "warm" to "cold" tier, providing a scalable and cost effective environment for data that is no longer required for active transactional processing but must be retained for compliance, auditing, or historical reference.
This tier houses archived case data, attachments, Business Intelligence Exchange (BIX) extracts, and system logs.
The financial motivation for transitioning data from the primary database to cold storage is profound, driven by the disparity in block-based pricing models and the hidden costs of database management. Total Cost of Ownership (TCO) analysis reveals that the primary database is one of the most expensive components of a SaaS subscription.
For example, SaaS products like Pega Cloud services in 2024 and 2025 demonstrate a clear pricing hierarchy. Primary Cloud Data Storage is typically provisioned in 100GB blocks, with each block carrying a monthly fee of approximately £200. In stark contrast, Cloud File Storage is provisioned in much larger 500GB blocks for a lower monthly fee of approximately £155.
By normalizing these figures to a price-per-gigabyte metric, the disparity becomes statistically significant:
Cloud Data Storage: ~₤ per GB per month
Cloud File Storage: ~₤0.31 per GB per month.
This represents an 84.5% reduction in direct storage costs for every gigabyte of data successfully migrated to the archival tier. For a large enterprise managing 20TB of historical cases, the annual savings from a robust archival strategy can exceed £400,000.
The economic burden of excessive database growth extends beyond the monthly storage fee. Large databases introduce several categories of "hidden" costs:
The most immediate benefit of data archival is often seen not in the balance sheet, but in the responsiveness of the application. The Pega Platform’s architecture is uniquely sensitive to database size due to its reliance on Rule-based logic and the storage of data in Binary Large Objects (BLOBs)
One of the biggest benefits of data archival is improved application performance. SaaS platforms such as Pegasystems rely heavily on database operations, and system performance can slow down significantly as historical data grows.
A key factor is the pzPVStream column, which stores case data in a compressed BLOB format. Every time a case is retrieved, the application must decompress the data and load it into JVM memory, increasing processing overhead.
The memory impact can be represented as:
JVM memory = BLOB size × decompression expansion factor + clipboard overhead
Research indicates that converting data for database storage can increase its size by up to 33% due to Base64 encoding and other serialization overheads. When millions of large cases remain in the primary database, the application must contend with a massive memory footprint for any operation that touches these historical records, such as global searches or comprehensive reporting.
Relational databases utilize B-tree indexes to accelerate data retrieval. As the number of rows in a table increases, the depth of these B-trees also increases. A query that requires three disk reads to traverse an index for a small table may require five or six reads for a table with hundreds of millions of rows. This results in a non-linear increase in query latency.
One common concern with archiving is that users may not be able to access old data easily. Most SaaS platforms now solve this by offering a read-only view of archived records, so users can still access historical information without affecting system performance.
While data remains accessible, there are technical constraints that must be understood during the design phase:
Determining the "better overall benefit" requires a nuanced assessment of the data’s utility. While archival offers superior economics and performance, there are specific scenarios where database persistence remains necessary.
Data should remain in the primary database if:
Archival is the optimal choice for:
Emerging technologies are beginning to use machine learning to predict when a case is likely to become "cold" even before it is officially resolved. By analyzing access patterns, these systems can pre-emptively move data to warmer tiers, optimizing costs in real-time without manual policy updates
Storage is no longer measured only in dollars, but in its carbon footprint. Cold storage, particularly tape-based or high-density HDD systems used by cloud providers for archival tiers, is significantly more energy-efficient than high-performance SSD-based relational databases.For organizations with ESG (Environmental, Social, and Governance) targets, moving data to the cloud archive is a quantifiable step toward reducing the environmental impact of their IT operations.
The analysis of SaaS data management shows one clear conclusion: moving inactive data to cold storage through a structured archival process is the best approach for most enterprises. This strategy helps organizations reduce costs, improve system performance, and meet compliance requirements more effectively.
Financially, the 85% reduction in storage costs provides an immediate and substantial Return on Investment (ROI). Technically, the reduction in memory pressure and the optimization of database query paths ensure that the application remains responsive and scalable as the business grows. Legally, the implementation of defensible disposal and immutable archival safeguards the organization against the mounting risks of data breaches and regulatory fines.
Sairohith Thummarakoti is an Enterprise Architect leading the design of Pega-based automation and decisioning systems for highly regulated industries including healthcare, banking, and insurance. His work focuses on scalable enterprise architecture, AI-driven workflow automation, cloud optimization, and intelligent data management for mission critical platforms. He currently serves as Chair of the IEEE Computer Society Columbia Section and is a selected speaker for the IEEE Computer Society Early Career Speaker Program (ECSP). Sairohith is also the author of seven books published through Taylor & Francis, has contributed more than 30 research papers, and regularly speaks at industry and academic conferences on AI, cloud computing, low-code/no-code platforms, and enterprise digital transformation.
Disclaimer: The authors are completely responsible for the content of this article. The opinions expressed are their own and do not represent IEEE’s position nor that of the Computer Society nor its Leadership.