
OUTLOOK 2025
There are obvious challenges related to parallelism. A major challenge for distributed networks is arriving at
exact consensus on the me-ordering of the transacon history. Simultaneous processing of transacons
may introduce dependencies at the transacon, block and smart contract level, requiring careful manage-
ment to maintain data order and consensus.
Parallelising a network might require even more powerful and expensive nodes, which hinders decentral-
isaon since the network will only be able to run on a smaller set of more deeply interconnected nodes.
New and ‘novel’ consensus mechanisms will also likely have to be invented to keep everything in sync. This
all has its own dangers and trade-os.
In short, achieving high performance and stability in blockchain networks is no small task, but teams are
paving the way forward with further parallelisaon of networks despite the inherent trade-os at hand. If
successful, these eorts could make blockchain a scalable and reliable technology, ready to support the
demands of tomorrow’s digital economy.
Source: Preda
OPTIMISTIC
PARALLELISATION
PESSIMISTIC
PARALLELISATION
ASYNCHRONOUS
PARALLELISATION
Examples Aptos, SEI v, Monad. Yes. Crystality (PREDA Model).
Execuon model Execute first, validate later. Pruned after ~ weeks. Decompose and execute
asynchronously with
non-overlapping state
dependencies.
Conict handling Detects conicts after
execuon and rolls back
conicng transacons.
Prevents conicts by locking
resources before execuon.
Nearly no rollbacks; No
stac analysis; Avoids nearly
all conicts by breaking
transacons into independent,
asynchronous steps.
Throughput High, if conicts are rare. Lower, due to locking and
reduced parallelism.
High concurrency by design,
with nearly no rollbacks
needed.
Concurrency High concurrency with
potential rollback costs.
Limited concurrency due to
locks.
Excellent, with a much higher
level of resource ulizaon.
Resource
ulizaon
Good, but can suffer due to
rollback overhead.
Poor, due to waing for locks. High, requires breaking
down transactions and
understanding async
programming.
Implementaon
complexity
Moderate, requires conflict
detection and rollback mech-
anisms.
Simple, but may suffer from
high contention management.
High, requires breaking
down transactions and
understanding async
programming.
Rollback
overhead
High in high-conflict scenarios. None, conflicts are prevented
before execution.
Nearly none, conflicts are
mostly avoided by decompos-
ing transactions.
Pros High throughput in low-con-
flict environments; Reduced
contention and waiting.
Ensures data integrity; Simple
to understand and implement.
High concurrency and scal-
ability; Nearly no rollbacks or
complex conflict resolution
needed.
Cons High cost of rollbacks in
high-conflict environments;
Complexity in conflict detec-
tion and resolution.
Reduced parallelism and
throughput; High contention
and waiting time.
Requires significant transac-
tion redesign; Steeper learning
curve for developers as it has
to redesign the programming
language.