Batch: Data is processed in batches.
Changes to records are typically collected over a period and then applied in bulk during scheduled batch jobs.
Batch processing typically involves
complex mechanisms (such as record-
locking) for synchronizing data across
different parts of the system, which
can lead to latency and potential data
Streaming: In FundGuard's streaming
architecture approach, data is processed and ingested as a continuous stream. Events are pushed to the system in real time, enabling near-instantaneous processing and analysis.
Data updates are processed as they
arrive, creating an immutable stream of
data records. Each update results in a
new version of the data.
Record Edits/Overrides: In this
model, records are mutable, meaning
they can be updated in place.
Changes are directly applied to the
existing records, which can lead to
issues like overwriting data or
inconsistencies during concurrent
This model uses less disk space due to
fewer records in tables and smaller
Historical data is either captured to a
warehouse at specific points in time,
or heavy processing is required to un-
process balances back to prior dates.
Back-dated changes can be difficult.
Record Immutability: In FundGuard's
streaming architecture, records are
immutable. Once a data entry is created,
it cannot be modified.
Any changes or updates are represented
as new entries with new versions to
balance tracking records. This
guarantees a clear audit trail and
simplifies data integrity.
Immutable streaming systems are
suitable for scenarios where data audit
trails, historical analysis, and real-time
analytics are important, such as
accounting data, audit logs, and
investment event-driven processes.
Controls & Rules
Controls Decoupled from Processes: Controls and exception rules are often
applied to batches of data before and
after bulk processing.
Data violations are only known at the
time of the bulk update and not when
data entries were made.
Control and rule evaluation might be
less granular, as batch update
designs tend to lower processing time
and use less data, rather than
Streaming Evaluations: Controls and
exception rules are applied as data
arrives in the stream.
Each data record is evaluated against
the rules in real time. Controls/rules are
applied at the record level, allowing for
rapid assessment and action.
Controls and rule evaluations tend to be
highly granular, as they are applied to
more detailed individual big-data
records or events as they are ingested.
Processing Time, Speed & Scale
Slower; Must Source Calculation Settings & Edit Records: Processing
speed will likely be slower compared
to streaming immutable systems due
to the relational searching nature of
configuration fields, and overhead of
record “edit” updates as opposed to
Latency is higher, and updates occur
in discrete intervals.
Faster; Record Adds; No Sourcing
Overhead: Processing is faster with
updates being record adds rather than
edits, and applied as soon as they are
received. Configuration settings are
stored at calculation points.
Processing updates are small and
frequent in contrast to traditional
systems that are mostly stationary and
waiting for bulk activities to execute.
Limits: Processing times can suffer
with high volumes of data as server
CPUs hit their maximums and
processing/database calculations are
less distributed horizontally.
Hardware specs must be based on
peak processing usage periods that
only occur at a very small period of
Changes in business operations (up or
down) can require large hardware
Near Endless: Streaming systems like
FundGuard are highly scalable,
distributing processing across multiple
nodes or microservices to handle high
Each microservice can process a specific
type of event, leading to flexibility on
processing power (elasticity) and helps
to alleviate high volume concerns.
Cloud-native frameworks like
FundGuard can harness the cloud's near
endless scalability, allowing client
organizations to quickly and easily scale
resources up or down as needed.
Testing & Debugging
Many Platform Components; Few Testing Tools: Testing record-edit
systems can lead to complex
Discovering what event(s) caused
incorrect rolling balances can be
difficult due to a lack of easy balance
This model tends to require large
testing efforts by operational staff
whenever new releases are introduced.
Getting platform components to work
together again after updating one or
more of them, can be time-
All In One Platform; Embedded Testing Tools: Testing event-driven
systems can be more complex than
Ensuring that events are correctly
processed using different sets of
configurations, and the system behaves
as expected, requires a specialized Test
Center where testing tools and
approaches let users load and reload
their own test scenarios with their
desired configurations, and includes an
auto verification of test results.
FundGuard’s single platform that
includes components like: file mapping,
exceptions, recon, report center, and
test center makes end to end
Data Privacy & Security
Security & Control via Isolation: It is
a common misconception that on-
premise systems offer more direct
control over data and therefore
But employing an on-prem approach
requires organizations to manage the
security of their physical
infrastructure, which is often solved
by database isolation and can have
negative impacts on resilience
measures and longer times-lapses in
disaster recovery situations.
More Secure; More Resiliency: Cloud-
native systems provide flexibility and
scalability with much more focus on
third-party access, and employ the best
of the public cloud provider's security
measures, availability zones and
Ensuring data privacy and security,
especially in multi-tenant environments,
includes robust access control
mechanisms and data at rest and data in
Brittle Integration Patterns: Often
rely on point-to-point integrations or
lack robust APIs, where custom code
or middleware connects directly to
specific third-party systems.
Often tightly coupled to specific
infrastructure, making it challenging to move or scale.
Active Integration Patterns: Cloud-
native applications like FundGuard favor
a more modular and decoupled
FundGuard uses API gateways, message
queues, and event-driven architectures
to integrate with third-party applications.
FundGuard is containerized,
orchestrated, and deployed
Narrow Observability: Monitoring
and tracing legacy integrations can be
challenging and customized.
Outdated security practices often
exist, such as hardcoding credentials.
Wide Observability: FundGuard
includes robust identity and access
management for secure
FundGuard is designed with monitoring
and observability in mind, making it
easier to collect and analyze metrics,
logs, and traces.