#IstioCon Is Your Virtual Machine Really Ready-to-go with Istio? Kailun Qin, Intel Haoyuan Ge #IstioCon Quick Summary (from Google Cloud Next ’19 [1]) VM works on Istio! [1] Istio Service Mesh Hybrid and Multi Clouds #IstioCon Istio VM Integration is? A Tumultuous Odyssey… [1] Istio 1.8: A Virtual Machine Integration Odyssey, Jimmy Song #IstioCon V0.2 Mesh Expansion ● Prerequisites ○ IP connectivity the workloads themselves #IstioCon V1.6-1.8 Better VM Workload Abstraction Item Kubernetes Virtual Machine Basic schedule unit Pod WorkloadEntry Component Deployment WorkloadGroup Service registry

#IstioCon Building resilient systems inside the mesh: abstraction and automation of Virtual Service generation Vladimir Georgiev, Thought Machine #IstioCon Sync calls failures inside the mesh implement this to be language agnostic? #IstioCon Virtual Services API ● Solves our problems, but… ● All Service Owners must be aware of the Virtual Services API in order to define their SLOs. ● Potential errors when dealing with YAMLs. ● Potential drift between the state of the service API and the Virtual Service config. ● Hard to manage when having hundreds of services. #IstioCon Abstracting to proto

will use cookie- based authentication mechanism bool ean 1.6.0 kyuubi.frontend.thr ift.http.cookie.dom ain Domain for the Kyuubi generated cookies strin g 1.6.0 kyuubi.frontend.thr ift r/docs/deployment/config/#yarn]): Name Default Meaning yarn.appmaster.vco res 1 The number of virtual cores (vcores) used by the JobManager (YARN application master). jobmanager.memor y.process.size docs Create a virtual environment named ‘kyuubi’ or anything you like using virtualenv if it’s not existing. New virtual environment Activate the virtual environment, Activate virtual environment Install

will use cookie- based authentication mechanism bool ean 1.6.0 kyuubi.frontend.thr ift.http.cookie.dom ain Domain for the Kyuubi generated cookies strin g 1.6.0 kyuubi.frontend.thr ift r/docs/deployment/config/#yarn]): Name Default Meaning yarn.appmaster.vco res 1 The number of virtual cores (vcores) used by the JobManager (YARN application master). jobmanager.memor y.process.size docs Create a virtual environment named ‘kyuubi’ or anything you like using virtualenv if it’s not existing. New virtual environment Activate the virtual environment, Activate virtual environment Install

will use cookie- based authentication mechanism bool ean 1.6.0 kyuubi.frontend.thr ift.http.cookie.dom ain Domain for the Kyuubi generated cookies strin g 1.6.0 kyuubi.frontend.thr ift r/docs/deployment/config/#yarn]): Name Default Meaning yarn.appmaster.vco res 1 The number of virtual cores (vcores) used by the JobManager (YARN application master). jobmanager.memor y.process.size docs Create a virtual environment named ‘kyuubi’ or anything you like using virtualenv if it’s not existing. New virtual environment Activate the virtual environment, Activate virtual environment Install

value denoted by d∗) α, β are dual feasible if α ⪰ 0, (α, β ) ∈ dom G and G > −∞ Often simplified by making implicit constraint (α, β ) ∈ dom G explicit Feng Li (SDU) SVM December 28, 2021 21 / 82 Weak �m i=0 αiy(i) = 0 The basic idea of SMO Algorithm 1 SMO algorithm 1: Given a starting point α ∈ dom J 2: repeat 3: Select some pair of αi and αj to update next (using a heuristic that tries to pick

Algorithm 1. The algorithm usually starts 2 Algorithm 1: Gradient Descent Given a starting point ✓ 2 dom J repeat 1. Calculate gradient rJ(✓); 2. Update ✓ ✓ � ↵rJ(✓) until convergence criterion is satisfied (7) Algorithm 2: Stochastic Gradient Descent for Linear Regression 1: Given a starting point ✓ 2 dom J 2: repeat 3: Randomly shu✏e the training data; 4: for i = 1, 2, · · · , m do 5: ✓ ✓ � ↵rJ(✓; x(i)

differentiable function using gradient descent Algorithm 1 Gradient Descent 1: Given a starting point θ ∈ dom J 2: repeat 3: Calculate gradient ∇J(θ); 4: Update θ ← θ − α∇J(θ) 5: until convergence criterion (Contd.) Algorithm 2 Stochastic Gradient Descent for Linear Regression 1: Given a starting point θ ∈ dom J 2: repeat 3: Randomly shuffle the training data; 4: for i = 1, 2, · · · , m do 5: θ ← θ − α∇J(θ;

will use cookie- based authentication mechanism bool ean 1.6.0 kyuubi.frontend.thr ift.http.cookie.dom ain Domain for the Kyuubi generated cookies strin g 1.6.0 Key Default Meaning Type pip install virtualenv Switch to the docs root directory. cd $KYUUBI_SOURCE_PATH/docs Create a virtual environment named ‘kyuubi’ or anything you like using virtualenv if it’s not existing. virtualenv

will use cookie- based authentication mechanism bool ean 1.6.0 kyuubi.frontend.thr ift.http.cookie.dom ain Domain for the Kyuubi generated cookies strin g 1.6.0 kyuubi.frontend.thr ift pip install virtualenv Switch to the docs root directory. cd $KYUUBI_SOURCE_PATH/docs Create a virtual environment named ‘kyuubi’ or anything you like using virtualenv if it’s not existing. virtualenv