A service interruption at Amazon Web Services (AWS) on October 20th impacted several Ethereum (ETH) wallets, including MetaMask, causing display issues and slowing down the Base network. This incident brought to light the vulnerabilities of decentralized systems when they rely heavily on a single cloud infrastructure provider.
AWS reported that the issue began in its US-EAST-1 region around 3:11 AM ET. Problems with DNS and EC2 load balancer health monitoring led to failures in DynamoDB and other associated services.
Amazon announced that the problem was significantly mitigated by 6:35 AM ET and fully resolved by evening, although some backlogs persisted into October 21st.
Coinbase acknowledged an active incident, stating that an “AWS outage impacting multiple apps and services” was affecting their platform. Simultaneously, users reported that their MetaMask balances were showing as zero and that transactions on the Base network were experiencing delays.
The technical dependency stems from Infura, which is the default Remote Procedure Call (RPC) provider for MetaMask. MetaMask documentation often directs users to Infura’s status page during such outages, as the wallet typically routes most read and write operations through Infura’s endpoints.
When Infura’s cloud infrastructure encounters problems, balance displays and transaction calls can report incorrect information. However, the funds themselves remain secure on the blockchain.
The AWS hiccup affected Ethereum and layer-2 networks utilizing Infura’s RPC infrastructure. This resulted in user interface (UI) errors that gave the impression of on-chain issues, even though the underlying consensus mechanisms were still functioning correctly.
Data from the Base chain on October 21st indicated a Total Value Locked (TVL) of $17.19 billion, roughly 11 million transactions within a 24-hour period, 842,000 active addresses daily, and $1.37 billion in Decentralized Exchange (DEX) volume over the preceding day.
Typically, brief outages lasting six hours or less lead to a decrease of 5% to 12% in DEX volume and a reduction of 3% to 8% in transaction counts. TVL usually remains stable as the problems are often superficial rather than fundamental.
More extended disruptions, lasting six to 24 hours, can cause a more substantial drop of 10% to 25% in DEX volume, an 8% to 20% decrease in transactions, and a 0.5% to 1.5% decrease in bridged TVL, as users delay bridging activities and shift to Layer 1 solutions as a risk mitigation strategy.
However, transaction volume and DEX volumes held fairly steady between October 20th and 21st. DEX volumes were $1.36 billion and $1.48 billion, respectively, while transactions totaled 10.9 million and 10.74 million, respectively.
The Base network also experienced a separate incident on October 10th involving delays related to high transaction volume, which was unrelated to AWS. The development team addressed and resolved the issue swiftly.
This prior incident underscored that layer-2 networks can encounter finality and latency issues during periods of high demand, irrespective of cloud infrastructure problems.
Combining demand-side pressures with external infrastructure failures amplifies the overall risk for networks that depend on centralized cloud providers.
| Date & time | Service | Update | Symptom | Resolved? |
|---|---|---|---|---|
| Oct 20, 07:11 | AWS (us-east-1) | Outage identified; internal DNS and EC2 load-balancer health-monitor fault | Global API/connectivity errors across major apps | “All 142 services restored” by 22:53; some backlogs lingered into Oct 21. |
| Oct 20, 07:28 → Oct 21, 00:57 | Coinbase status | Incident opened → resolved | Users unable to login, trade, transfer; “funds are safe” messaging | Recovered; monitoring through evening Oct 20 (PDT). |
| Oct 20, 19:46 | Decrypt tracker | MetaMask balances showing zero; Base/OpenSea struggling as AWS issues persist; Infura implicated | Wallet UI misreads and RPC errors across ETH & L2s | Ongoing during afternoon; recovery staggered by provider queues. |
| Oct 10, 21:40 (context) | Base status | “Safe head delay from high tx volume” (unrelated to AWS) | Finality/latency lag (“safe head” behind) | Resolved same day; shows L2 latency edges independent of cloud events. |
Cloud concentration surfaces as a systemic weakness
This AWS event brings renewed attention to ongoing concerns regarding the concentration of crypto infrastructure within a small number of cloud providers.
Previous AWS incidents in 2020, 2021, and 2023 revealed intricate interdependencies among services like DNS, Kinesis, Lambda, and DynamoDB. These interdependencies propagate to wallet RPC endpoints and layer-2 sequencers hosted on the cloud.
Because MetaMask defaults to routing through Infura, even a minor cloud issue can appear to affect the entire chain for end-users, regardless of whether on-chain consensus is functioning normally.
The Solana network’s five-hour halt in 2024, which was caused by a software flaw, demonstrated that users are willing to tolerate short periods of downtime if the recovery process is well-managed and communication is transparent.
Optimism and Base have previously encountered “unsafe” and “safe” head stalls on their OP-stack architecture, but the development teams have generally resolved these issues through protocol improvements.
The recent AWS disruption is different, in that it highlights infrastructure dependencies existing outside of the blockchain protocols themselves.
As a result of this incident, infrastructure teams are likely to accelerate plans for multi-cloud failover and expand the diversity of RPC endpoints.
Wallets might start prompting users to configure custom RPCs rather than relying solely on a single default provider.
Layer-2 teams usually publish post-mortem reports and revisions to their service-level objectives within one to four weeks after major incidents. This may lead to increased prioritization of client diversity and multi-region deployments in upcoming roadmaps.
What to watch
AWS is expected to release a post-event summary that details the root causes of the US-EAST-1 disruption and the steps taken to remediate it.
The Base and Optimism teams are also expected to publish incident post-mortems addressing any impact on sequencers or RPC services specific to OP-stack chains.
RPC providers, including Infura, are facing pressure to make public commitments to multi-cloud architectures and geographic redundancy. This redundancy is critical for withstanding failures at a single cloud provider.
Centralized exchanges that experienced incidents during the AWS outage, including Coinbase, may see a widening of spreads and a shift in trading volume to decentralized exchanges on less affected chains during future cloud disruptions.
Monitoring exchange status pages and downdetector curves during infrastructure events provides real-time insight into how centralized and decentralized trading venues perform differently under stress.
The event reinforces the fact that blockchain’s decentralized consensus cannot fully shield the user experience from bottlenecks in centralized infrastructure.
The concentration at the RPC layer remains a practical point of weakness, where cloud provider failures manifest as wallet display errors and transaction delays. These issues undermine confidence in the reliability of Ethereum and layer-2 ecosystems.
