Walrus ($WAL ) change la conversation sur le stockage de "qui détient les fichiers" à "comment efficacement le réseau peut les garder récupérables." Les anciens systèmes de stockage décentralisé s'appuient souvent sur une réplication lourde ou une permanence coûteuse, ce qui peut rendre la fiabilité réelle mais les coûts difficiles à contrôler. Walrus adopte une approche plus axée sur l'ingénierie : diviser les données, les encoder pour la récupération, les distribuer à travers les nœuds et payer pour la performance plutôt que pour des promesses. Cela compte car les bâtisseurs n'ont pas seulement besoin que les données existent, ils ont besoin qu'elles restent récupérables sous charge, fluctuation et mauvaises conditions de réseau. Si le stockage est la fondation, l'efficacité devient la caractéristique qui décide de l'adoption.@Walrus 🦭/acc #walrus $WAL
Walrus ($WAL ) n'essaie pas de "décentraliser le stockage" comme un slogan. Il essaie de supprimer le mode de défaillance silencieuse qui brise les vraies applications : des données qui semblent disponibles mais qui ne sont pas fiables lorsqu'il est important de les récupérer. La plupart des produits Web3 dépendent encore des liens cloud, des passerelles ou des chemins d'indexation fragiles, donc les utilisateurs finissent par faire confiance à des intermédiaires au lieu d'une infrastructure. Walrus considère le stockage comme un service au niveau du protocole, où la disponibilité est conçue par l'encodage, la distribution et des incitations qui récompensent la capacité honnête et le temps de disponibilité. Le résultat est simple mais important : moins d'états inconnus, des garanties plus claires, et une couche de stockage qui semble suffisamment stable pour construire des produits sérieux.@Walrus 🦭/acc #walrus $WAL
Dusk est construit autour de l'idée que la sécurité n'est pas seulement du code, c'est un comportement sous pression. Le staking transforme les validateurs en opérateurs responsables en rendant l'honnêteté le chemin rentable et les échecs coûteux. Lorsque les validateurs s'engagent à staker, ils ne gagnent pas seulement des récompenses, ils prennent la responsabilité de la disponibilité, du vote correct et de la stabilité du réseau. Cela compte pour le règlement financier, où la chaîne ne peut pas se permettre une fiabilité "meilleur effort". Les incitations sont le niveau d'application : une bonne performance vous garde dans le jeu, une mauvaise performance vous coûte. La conception de Dusk traite la sécurité comme un budget opérationnel, pas un slogan.@Dusk #dusk $DUSK
Dusk built Piecrust because regulated finance needs smart contracts that can use cryptography as a first-class tool, not as an awkward add-on. Most VMs were designed for general apps, so anything advanced like proofs, signatures, or confidential logic becomes heavy, slow, or outsourced to external systems. Piecrust is designed around crypto-native host functions, meaning contracts can call efficient built-in primitives instead of reinventing them every time. That keeps execution cleaner and more predictable, especially for workflows that depend on privacy and controlled disclosure. If Dusk is aiming for real markets, its VM stack has to match that reality.@Dusk #dusk $DUSK
Dusk utilise deux modèles de transaction car la finance réelle nécessite deux modes : l'utilisabilité et la confidentialité. Moonlight est la voie basée sur les comptes qui semble familière pour les transferts normaux et les flux de travail d'applications, où la transparence est acceptable et l'intégration est simple. Phoenix est la voie UTXO conçue pour la confidentialité, où la valeur se déplace comme de la monnaie numérique et où les activités sensibles peuvent rester cachées des observateurs publics. Le but n'est pas "la confidentialité pour cacher", c'est la confidentialité pour réduire les fuites d'informations sur les marchés sérieux. Une chaîne, deux rails, choisis par cas d'utilisation. Ce design rend Dusk plus pratique que les réseaux qui forcent tout dans un modèle unique.@Dusk #dusk $DUSK
Dusk ($DUSK ) ressemble à l'une des rares chaînes construites pour de véritables flux de travail financiers, pas seulement pour le trading de cryptomonnaies. Dans les marchés, la confidentialité ne consiste pas à disparaître, mais à protéger l'exécution sensible tout en gardant la capacité de prouver des choses lorsque cela est nécessaire. C'est ce que Dusk vise : des transactions confidentielles avec une structure qui peut encore soutenir la conformité et la divulgation contrôlée. Cela compte pour les institutions car le plus grand risque n'est pas la volatilité, mais l'incertitude des règles, des rapports et des résultats de règlement. Si la finance sur chaîne doit se développer au-delà des expériences, c'est ce type de conception dont elle a besoin.@Dusk #dusk $DUSK
Dusk ($DUSK ) est conçu pour un monde où la finance a besoin de confidentialité, mais où les institutions et les régulateurs ont toujours besoin de clarté. La plupart des chaînes traitent la confidentialité comme "cacher tout", ce qui semble bien jusqu'à ce que le capital réel arrive et que les exigences d'audit se manifestent. Dusk adopte une approche plus stricte : les transactions peuvent rester confidentielles, tandis que le système prend en charge la divulgation contrôlée lorsque les règles l'exigent. Cela compte pour les titres tokenisés, l'exécution de style OTC, et tout marché où les détails de la transaction ne peuvent pas fuiter sans causer de dommages. L'objectif n'est pas le secret. Il s'agit de réduire l'exposition, de prévenir la dérive de l'information, et de maintenir la confiance dans le règlement.@Dusk #dusk $DUSK
BTC 100K $ : Pas une Question de "Si" — Mais une Question de Carburant
Le prix de 100 000 $ pour le Bitcoin n'est plus un coup de chance marginal ; c'est un objectif structurel intégré dans le paysage financier actuel. La question n'est pas si, mais quoi fournira l'allumage final. Une analyse plus approfondie révèle une convergence critique de trois nouvelles sources de carburant, allant au-delà des simples récits d'ETF.
Carburant 1 : Le "Cycle de Recarburage" Institutionnel Les ETFs Spot ne sont pas juste un afflux ponctuel ; ils ont créé une nouvelle mécanique de marché. Plus de 90 % de leurs achats sont "non discrétionnaires", dictés par la création/destruction de fonds. Cela crée une offre constante et algorithmique pour du BTC physique, directement à partir de l'offre circulante. Chaque dollar investi n'est pas juste un sentiment — c'est une réduction permanente des pièces disponibles, resserrant le vise quotidiennement.
From Tokenization to Settlement: Why Dusk Targets Real Financial Workflows
Most crypto projects talk about “tokenization” like it ends at minting an asset on-chain. But real finance does not stop at issuance. The hard part starts after that: rules, transfers, settlement, privacy, reporting, and the daily reality of operating in regulated markets. Dusk is designed around that full lifecycle. It positions itself as a Layer-1 built for regulated finance, where assets can be issued, moved, and settled with privacy and compliance built into the system, not bolted on later. Tokenization is easy. Running a market is not. If you issue a token that represents a security, bond, or real-world instrument, you immediately inherit real obligations. Who is allowed to hold it? Can it be transferred freely? Are there limits, restrictions, eligibility checks, and disclosures? If something goes wrong, who can prove what happened? Most general-purpose blockchains were built to be open and transparent. That is great for public DeFi, but it creates two problems for regulated assets. First, everything is visible, which means traders, market makers, and institutions leak sensitive data every time they interact. Second, compliance becomes an off-chain headache, where rules live in spreadsheets, legal documents, and centralized backend logic that users cannot verify. Dusk is aiming to bring those constraints on-chain in a way that feels native to financial systems. Privacy in finance is operational, not ideological There is a difference between “privacy coins” and financial privacy. Institutions are not trying to disappear. They want to protect execution details, positions, counterparties, and strategy. In markets, leaking that information is not a small issue, it is direct risk. Dusk’s approach is privacy by design, with the ability to be transparent when needed. That phrase matters. It implies controlled disclosure instead of permanent exposure or total secrecy. This is one reason Dusk supports two transaction models inside its core transfer system: Moonlight (account-based public transfers) and Phoenix (UTXO-based shielded transfers). It gives builders flexibility to choose the right tool for the right part of a workflow. Settlement finality is the real product In regulated finance, the most expensive word is “pending.” It creates disputes, delays, and operational stress. Dusk puts settlement finality at the center of its design using its Proof-of-Stake consensus called Succinct Attestation, described as a committee-based system built for deterministic finality suitable for financial markets. That matters because tokenization without reliable settlement is just digital paperwork. Real markets need the system to say: this transfer is final, irreversible, and verifiable. Only then can firms build serious workflows on top without living in exception handling mode. Modular design: execution can change, settlement cannot Another practical decision is Dusk’s modular architecture: it separates the settlement layer (DuskDS) from execution environments like DuskEVM. The point is not to chase hype about compatibility. The point is to keep the base layer focused on the boring, critical job: consensus, data availability, and settlement integrity. Execution environments evolve. Requirements change. But regulated markets care about the settlement engine being stable and predictable. Why this direction is more realistic than “one chain for everything” Dusk is not trying to win by being the fastest general chain or the loudest ecosystem. It is trying to be the infrastructure that regulated markets can actually adopt without breaking their own rules. That means building for: confidential transfers where exposure is dangerous, compliant flows where eligibility and disclosure matter, and deterministic settlement where “final” is non-negotiable. Tokenization is the headline. Settlement is the business. Dusk is building for the part that institutions cannot compromise on. @Dusk #dusk $DUSK
MiCA Reality Check: What Regulation Forces Crypto to Fix (And Why Dusk Was Built for This World)
For years, crypto has moved fast mainly because the rules were unclear. That era is ending in Europe. The Markets in Crypto-Assets Regulation (MiCA) is not a small policy update, it is a framework designed to bring order to how crypto is issued, marketed, and offered as a service across the EU. The message is simple: if crypto wants real adoption, it cannot live in permanent “beta mode.” MiCA targets two big areas. First: public offers of crypto-assets, meaning how tokens are launched, sold, and disclosed to users. Second: crypto-asset service providers (CASPs) like exchanges, custodians, and brokers, meaning how platforms operate, handle assets, and protect customers. It pushes for market integrity, transparency, and consumer protection, and regulators are already warning companies not to misuse the “regulated” label as marketing. Now here’s the problem: most blockchains were not designed for this environment. Public-by-default ledgers are great for transparency, but they create a new risk for serious financial use. If every trade, balance, and flow is visible, it becomes a data leak. Institutions cannot expose execution details and positions to the entire internet. That is not “decentralization.” That is forced disclosure. Dusk’s core thesis is that without privacy, institutional adoption stays out of reach, but privacy must still allow compliance when required. This is where Dusk becomes interesting in the MiCA era. Dusk positions itself as a privacy blockchain for regulated finance, meaning it is not treating privacy like hiding. It treats privacy like a controlled system: users and institutions can operate with confidentiality, while the network can still support regulatory requirements and audits under clear rules. The key shift is this: Privacy cannot be an “escape hatch.” It has to be part of the operating model. And it must be compatible with rules. Dusk’s documentation is direct about that goal, and it’s also reflected in how the protocol is structured. It includes components designed for regulated markets, like Zedger, which targets tokenized securities use cases where privacy and compliance are both non-negotiable. MiCA also highlights why stablecoins and settlement infrastructure will face more scrutiny. EU regulators are actively discussing stablecoin risks, including redemption rights and multi-issuance models, because stablecoins behave like money in practice. That means the settlement layer behind them needs real finality, governance clarity, and predictable operations, not vague “it should confirm soon” behavior. Dusk’s approach is built around that reality: regulated markets need final settlement, not just activity. Dusk emphasizes performance and finality to meet the demands of financial workflows, and its updated whitepaper messaging frames the chain as “privacy-focused” and “compliance-ready” specifically for regulated finance. What MiCA really forces crypto to fix is not one thing, it is the entire mindset. Projects need clearer disclosures. Service providers need stronger operational discipline. And the industry needs to stop pretending that “transparent by default” is always a feature. In regulated finance, transparency must be intentional, not automatic. That is why Dusk’s positioning makes sense in a post-MiCA world. It is aiming for a future where confidentiality protects market participants, while controlled auditability keeps the system usable for real institutions. Not privacy instead of compliance, but privacy as a compliance tool. If crypto is becoming regulated infrastructure, then Dusk is building for the version of crypto that has to survive that upgrade. @Dusk #dusk $DUSK
Moonlight vs Phoenix: Why Dusk Uses Two Transaction Models (And Why That Matters)
Most blockchains pick one transaction model and commit to it forever. Dusk does something more practical: it runs two models side by side, because privacy and usability do not always want the same architecture. On Dusk, transactions are handled by the Transfer Contract, which supports both a transparent account-based flow (Moonlight) and a confidential UTXO-based flow (Phoenix). That dual setup is not marketing. It is an engineering decision to balance “easy payments + smart contract UX” with “strong privacy for sensitive finance.” Moonlight: the “normal” account-based lane Moonlight is the side of Dusk that feels familiar to most users. Account-based systems behave like bank balances: you have an account, it has a number, and that number goes up or down as you send and receive. This model is usually simpler for everyday activity. Wallets can show balances clearly, gas payments are straightforward, and building apps often feels more direct because state lives in accounts. For most Web3 users, this is the mental model they already understand. Moonlight exists so Dusk can support smoother experiences where full privacy is not the priority, such as regular transfers, public actions, or workflows where being transparent is fine. Phoenix: the private UTXO lane Phoenix is where Dusk gets serious about privacy. Phoenix uses a UTXO-style architecture, which is closer to Bitcoin’s model: instead of “balances,” you spend discrete outputs like digital cash notes. Why does that matter? Because UTXO models can be better for privacy and parallel spending design, especially when paired with cryptography. Phoenix is designed for obfuscated/confidential transactions and even confidential smart contract activity in a way that does not leak the same kind of obvious account trails. Dusk explains Phoenix outputs as being stored in a Merkle tree, and users spend them by proving knowledge of the path and commitment opening, rather than exposing everything in plain sight. This is the foundation that enables confidential movement without turning the chain into a black box that institutions cannot touch. And importantly, Phoenix is not just “privacy vibes.” Dusk published updates showing Phoenix has formal security proofs, positioning it as a serious transaction model rather than an experimental privacy add-on. Why Dusk keeps both instead of choosing one The simple answer: finance has multiple modes. Some actions need to be transparent and easy. Others need to be private because the cost of leaking data is real. In trading, settlement, and issuance workflows, information leakage can become a direct disadvantage. Many institutions will not touch on-chain systems if every action exposes positions, counterparties, or execution details. Dusk’s dual design lets builders choose the right lane: Moonlight for speed, UX, and normal contract interaction. Phoenix for confidentiality where it actually matters. This is why the docs frame Dusk transactions as both “transparent and obfuscated,” managed under one protocol-level transfer system. The compliance angle: private by default, but not unusable A lot of privacy chains fail because they treat privacy like total invisibility. That creates problems for regulated markets, because regulated markets do not just want secrecy, they want controlled disclosure and accountable settlement. Dusk’s approach is closer to real-world finance: keep sensitive details private, while still building a system that can support rules and enforcement where required. The network is explicitly designed for fast, deterministic final settlement, which is a requirement in financial workflows. Phoenix handles confidentiality. Moonlight supports usability. And the chain’s settlement layer focuses on finality that markets can rely on. The real takeaway for builders and investors Moonlight vs Phoenix is not “two features.” It is a statement about what Dusk is trying to become. Dusk is building a chain where privacy is not a loophole, it is a core market primitive. At the same time, it avoids the trap of making everything private, slow, or hard to use. In practice, this dual-transaction model is what makes Dusk more than just another L1. It is trying to be a place where you can run financial activity that is confidential when needed, transparent when it should be, and final when it must be. That is the kind of system institutions can actually plug into, and builders can actually ship on. @Dusk #dusk $DUSK
Walrus : Quand la disponibilité des données est devenue un protocole, pas une promesse
Tout a commencé par une facture que je ne m'attendais pas. Notre équipe venait de lancer une nouvelle fonctionnalité qui permettait aux utilisateurs de télécharger de courts clips et des images. Rien de fou. Quelques secondes de vidéo, quelques captures d'écran, quelques métadonnées. L'application progressait rapidement, la croissance était régulière, et tout semblait « bien » jusqu'à ce que la facture de stockage arrive. Ce qui me faisait peur, ce n'était pas le chiffre en lui-même. C'était le schéma. Chaque nouvel utilisateur devenait discrètement un coût à long terme. Le stockage centralisé est simple quand vous êtes petit. Vous choisissez Amazon S3 ou Google Cloud, branchez un seau et passez à autre chose. Mais une fois que votre produit dépend des médias, vous arrêtez de payer pour le « stockage ». Vous commencez à payer pour l'ensemble du système qui l'entoure : les sorties, la réplication, les zones de disponibilité, les pics de bande passante et tout changement de politique qui pourrait survenir le trimestre prochain. Cela devient une dépendance avec laquelle vous ne pouvez pas négocier.
#walrus is trying to solve a real problem most people ignore: data control quietly centralizes as networks scale. It is easy to launch a “decentralized” system, but as usage grows, power can concentrate in a few large operators unless the incentives fight back. Walrus focuses on keeping storage decentralized by design, not by marketing. Stake delegation spreads power across independent storage nodes so no single party controls what gets stored or censored. Rewards are tied to verifiable uptime and reliability, so smaller operators can still compete if they perform well. If nodes act dishonest or unreliable, they lose stake, which makes bad behavior expensive. Walrus also adds friction to fast stake movements to reduce coordinated power grabs during sensitive moments like votes or attacks. With Seal, developers can even build programmable access control, so privacy can be enforced at the data layer, not just promised. In short: Walrus scales by making decentralization economically rational. @Walrus 🦭/acc $WAL
Walrus essaie de résoudre la partie de l'automatisation de l'IA qui semble encore peu sûre : laisser un agent dépenser de l'argent sans transformer votre compte bancaire en un jeu de devinettes. Les “paiements agentiques” ne fonctionnent que lorsque vous pouvez prouver ce que l'agent a vu, quelles règles il a suivies, et pourquoi il a pris cette décision. Sinon, vous faites confiance à une boîte noire à 2 heures du matin avec de réelles conséquences financières. Walrus rend la mémoire de l'agent vérifiable par défaut, avec une preuve de disponibilité et des métadonnées ancrées sur Sui, de sorte que les entrées derrière un achat peuvent être vérifiées plus tard au lieu d'être discutées. Cela transforme les actions en pistes d'audit, pas en impressions. Ajoutez un contrôle d'accès cryptographique pour les données de paiement sensibles, et vous obtiendrez un modèle plus propre : les agents peuvent agir rapidement, mais les humains peuvent toujours inspecter, limiter et vérifier ce qui s'est passé. @Walrus 🦭/acc #walrus $WAL
WAL’s Real Job: Turning Storage Into a Paid Reliability Service
Walrus is not trying to be “another storage network.” It is trying to fix a problem most Web3 builders quietly accept: your app can be on-chain, but your real data usually isn’t. Images, user uploads, game files, AI datasets, and app history often sit on normal cloud storage because it’s easy, fast, and familiar. But that choice brings a hidden cost. You don’t just rent storage, you rent rules. Pricing can change, access can be limited, and policy decisions can break your product overnight. Walrus is built for builders who want an exit. Not from the cloud’s performance, but from its control. Not from convenience, but from dependence. The WAL token sits at the center of that model, but it’s important to understand what it actually represents. In most crypto projects, a token is either a fee coupon or a governance badge. WAL is closer to an operational unit: it coordinates storage service at the protocol level and connects payment to real reliability. That sounds simple, but it changes how storage behaves in a decentralized system. Traditional cloud storage works because one company controls the whole pipeline. They own the servers, route traffic, manage failures, and enforce consistency. When you pay them, you are paying for that coordination. Decentralized storage struggles because coordination is the hard part. If storage is spread across many independent nodes, the network must still answer one question cleanly: “Is this data available right now, and will it stay available?” Walrus tries to solve that with a design that treats storage as a single service, even though it is delivered by many nodes. Files are split into pieces and distributed across a network so that the system can recover data even if some nodes go offline. The goal is not just survival in worst-case scenarios. The goal is stable availability under normal daily conditions, where nodes churn, networks slow down, and workloads spike. Here is where WAL matters. Walrus does not want users to negotiate storage like a marketplace transaction every time. That would create friction and uncertainty. Instead, WAL supports a coordinated model where users pay for storage in a predictable way, and the network assigns work across nodes based on capacity and performance incentives. In other words, WAL is part of a system that pays nodes to behave like infrastructure, not like opportunistic sellers. A key idea is that reliability should be rewarded, and unreliability should be expensive. This sounds obvious, but many decentralized storage systems fail here. They pay for “being part of the network” instead of paying for “being good at the job.” Walrus leans toward performance-linked incentives. Nodes that deliver consistent availability and meet commitments are the ones that should earn more. Nodes that over-promise, under-deliver, or disappear should lose out. That incentive shape matters because it changes node behavior. If failure has no real cost, nodes can accept more load than they can handle, hoping they still get paid. If failure is punished, nodes must commit realistically. That pushes the network toward honest capacity reporting. It also reduces the risk that reliability becomes a marketing slogan instead of a measurable outcome. Another part people miss is that storage is not just “space on disk.” Writes cost work. A decentralized network must encode data, distribute it, verify pieces reached the right places, and keep proofs or confirmations that it is still retrievable. That consumes bandwidth and compute. If a pricing model only charges for “keeping data,” it ignores the heavier part of the job: safely onboarding data into the network and maintaining its recoverability guarantees. Walrus pricing recognizes this by treating storage as a full lifecycle service. You are not just paying for bytes sitting somewhere. You are paying for the network to do the operational work to make those bytes durable and available across time. WAL becomes the unit that aligns that cost with the people doing the work. This is also why Walrus fits the Sui ecosystem story well. Sui is optimized for fast execution and modern app designs. But apps don’t live on transactions alone. They live on content, state history, media, and user-generated data. If that layer remains centralized, the whole app becomes “partly Web2” no matter how on-chain the smart contract logic is. Walrus is trying to remove that weak link by giving builders a storage layer that feels like a real service, not a fragile set of peers. For creators and investors, the clean way to look at WAL is not “what number goes up.” It is: does this token represent demand for storage reliability, and does the system turn that demand into sustainable node economics? If Walrus succeeds, WAL is tied to real usage, because real apps pay for storage. That makes it different from tokens that only circulate around speculation. In practice, this can lead to a healthier product loop. More app usage means more storage demand. More storage demand means more node revenue. More node revenue attracts more capacity and better operators. Better operators improve performance and reliability, which makes the storage layer easier to trust. And when a builder can trust storage, they can ship faster, reduce incident risk, and stop building complicated workarounds to protect themselves from centralized lock-in. Walrus isn’t promising magic. It’s promising something more valuable: a boring, dependable storage foundation that doesn’t require permission from a single gatekeeper. WAL is the mechanism that makes that foundation pay for itself. @Walrus 🦭/acc #walrus $WAL
Plus de 'Probablement Bien' : Comment Walrus Élimine la Zone Grise dans le Stockage
Ce n'était pas en panne. Cela ne fonctionnait pas. C'était juste flou. Cette zone 'floue' est là où la plupart des douleurs de stockage résident. Pas dans des pannes dramatiques, pas dans des échecs bruyants, mais dans ces longues nuits où tout semble principalement normal et rien n'est vraiment fiable. Un fichier semble accessible. Un récupérateur rapporte un succès. Un indexeur affirme que les données sont présentes. Et pourtant, l'expérience utilisateur dit le contraire.
C'est le pire genre d'échec car il refuse d'être mesuré clairement. Quand quelque chose est complètement en panne, les gens réagissent rapidement. Les alertes se déclenchent. Les incidents sont déclarés. Tout le monde sait ce qui doit se passer ensuite. Mais quand un système reste à moitié correct, les équipes perdent des heures à poursuivre des ombres. Vous ne réparez pas une pièce cassée. Vous essayez de prouver ce qui est réel.
Protocole Walrus : Stockage Programmable que l'IA et le Web3 peuvent réellement construire
Les données sont le véritable carburant. Le stockage est le véritable goulot d'étranglement. Le morse essaie de réparer les deux. La plupart des gens pensent que le Web3 concerne uniquement les contrats intelligents et les chaînes plus rapides. Mais quand on regarde de plus près, de nombreuses applications dépendent encore des mêmes anciennes fondations. Les images NFT, les fichiers de jeu, les publications sociales, et même les ensembles de données d'IA vivent souvent sur un stockage cloud normal. Cela rend l'expérience fragile. La chaîne peut être décentralisée, mais l'application elle-même a toujours un point faible central.
Le protocole Walrus est conçu pour résoudre ce problème. Il se concentre sur le stockage de grandes quantités de données de manière décentralisée, tout en les rendant utilisables pour des applications réelles. L'idée principale est simple : les données ne devraient pas être coincées à l'intérieur des serveurs d'une seule entreprise. Elles devraient être distribuées à travers de nombreux nœuds indépendants, afin qu'elles restent disponibles même lorsque certaines parties du réseau sont hors ligne.